Click here to go to the program overview page.
Click here if you are an author for access to submissions.
Further details will be updated here.


Hide contributed papers.
Hide poster highlights.

Click here to see the Thursday program on PaperCept.

Thursday, 22 Aug

Keynote by Peter Stone

Thursday, 22 Aug, 9:00am-9:45am
Introduced by Lorenzo Sabbatini
Advances in Ad Hoc Teamwork and Adaptive Tolling for Multiagent Traffic Optimization

As autonomous agents proliferate in the real world, both in software and robotic settings, they will increasingly need to band together for cooperative activities with previously unfamiliar teammates. In such “ad hoc” team settings, team strategies cannot be developed a priori. Rather, an agent must be prepared to cooperate with many types of teammates: it must collaborate without pre-coordination. This talk begins by presenting recent advances in ad hoc teamwork. The second part of the talk introduces Delta-tolling, a novel adaptive pricing scheme for multiagent traffic optimization.

Poster Highlights I

Thursday, 22 Aug, 9:45am-10am
Session Chair: Kiril Solovey

ROS-CBT: Communication Benchmarking Tool for the Robot Operating System
Masaba, Kizito;  Quattrini Li, Alberto
This paper presents the Robot Operating System Communication Benchmarking Tool (ROS-CBT). This tool addresses the limitation in ROS of simulating communication between robots in existing robot simulators, such as Stage and Gazebo. Specifically, ROS-CBT, a ROS package, emulates areal communication link between any two robots on a network in a transparent way – that is, the ROS framework and existing packages do not need to be modified, allowing for easy porting of the robotic software to real robots. It also provides features for performance evaluation of each communication link at runtime. Finally, we discuss on lessons learned and possible future directions for the support of multi-robot systems within the ROS framework. Experiments and results show ROS-CBT applicability to real-time analysis of relevant multi-robot applications, which contributes to accelerate test and development of multi-robot systems.

Experimental Validation of Stable Coordination for Multi-Robot Systems with Limited Fields of View Using a Portable Multi-Robot Testbed
Mukherjee, Pratik;  Santilli, Matteo;  Gasparri, Andrea;  Williams, Ryan
In this paper, we address the problem of stable coordinated motion in multi-robot systems with limited fields of view (FOVs). These problems arise naturally for multi-robot systems that interact based on sensing, such as our case study of multiple unmanned aerial vehicles (UAVs) each equipped with several cameras that are used for detecting neighboring UAVs. In this context, our contributions are: i) first, we derive a framework for studying stable motion and distributed topology control for multi-robot systems with limited FOVs;  and ii) Then, we provide experimental results in indoor and challenging outdoor environments (e.g., with wind speeds up to 10 mph) with a team of UAVs to demonstrate the performance of the proposed control framework using a portable multi-robot experimental set-up.

A Hardware and Software Testbed for Underactuated Self-Assembling Robots
Nilles, Alexandra;  Wasserman, Justin;  Born, Austin;  Horn, Chris;  Born, John;  LaValle, Steven M
We present the implementation and characterization of an affordable testbed for underactuated multi-agent, self-assembling systems. There has been recent interest into the control of nano- and micro-scale active particle systems, but these systems are often difficult to manufacture and observe, hindering control research. Our testbed offers an accessible way to experiment with different design and control approaches. The testbed is composed of an off-the-shelf rolling weaselball toy and a 3D printed external hub that modifies the agent’s dynamics. The software toolbox includes simulations and code for data extraction and analysis of the weaselballs. The advantage of our testbed for studying distributed robotic systems is that these robots can be made quickly and cheaply, are relatively small, and do not require complex or expensive environments. The software in our toolbox includes a high fidelity Gazebo simulation, and Python code for collecting trajectories and other data from both simulation and overhead video of the system. Using this toolbox, we present useful computed properties of the system with regards to object clustering.

Effective Heuristics for Multi-Robot Path Planning in Warehouse Environments
Han, Shuai D.;  Yu, Jingjin
In this preliminary study, we propose a new centralized decoupled algorithm for solving one-shot and dynamic optimal multi-robot path planning problems in a grid-based setting mainly targeting warehouse like environments. In particular, we exploit two novel and effective heuristics: path diversification and optimal sub-problem solution databases. Preliminary evaluation efforts demonstrate that our method achieves promising scalability and good solution optimality.

LLDM: Locally Linear Distance Maps for Robot Motion Planning
Putman, Josiah;  Oh, Lisa;  Zhao, Luyang;  Honnold, Evan;  Brown, Galen;  Wang, Weifu;  Balkcom, Devin
This paper presents a data structure that summarizes distances between configurations across a robot configuration space, using a binary space partition whose cells contain parameters used for a locally linear approximation of the distance function. Querying the data structure is extremely fast, particularly when compared to graph search required for querying Probabilistic Roadmaps, and memory requirements are promising. The paper explores the use of the data structure constructed for a single robot to provide a heuristic for challenging multi-robot motion planning problems. Potential applications also include the use of remote computation to analyze the space of robot motions, which then might be transmitted on-demand to robots with fewer computational resources.

Paper Presentation I: Planning and task allocation

Thursday, 22 Aug, 10am-11am
Session Chair: Michael Otte

On Minimum Time Multi-Robot Planning with Guarantees on the Total Collected Reward
Sadeghi, Armin;   Asghar, Ahmad Bilal;   Smith, Stephen L.
In this paper, we study a multi-robot planning problem where a team of robots visit locations in an environment to collect a specified amount of reward in the minimum possible time. Each location has a weight associated with it that represents the value or amount of reward that can be collected by visiting that location. The problem is to design tours for the robots to collect at least D units of reward while minimizing the length of the longest robot tour. The single robot unweighted version of this problem is called the k-Stroll problem. We provide a 3-approximation algorithm for the multiple robot version of the k-Stroll problem. This leads to an algorithm for the weighted problem that collects at least (1-e)D reward with tour lengths of at most 3 times the optimal tour length. The analysis of the approximation algorithm is then extended to provide bi-criterion approximations to two variations of the problem. We provide an application of the approximation algorithm for planning UAV tours with gimballed cameras for monitoring an urban environment.

Decentralized Multiple Mobile Depots Route Planning for Replenishing Persistent Surveillance Robots
Ding, Yifan;   Luo, Wenhao;   Sycara, Katia
Persistent surveillance of a target space using multiple robots has numerous applications. The continuous operation in these applications is challenged by limited onboard battery capacity of the persistent robots. We consider the problem for replenishing persistent robots using mobile depots, where mobile depots collectively compute a set of tours to drop off batteries for replenishing all persistent robots with the minimum total cost. Compared to other works, persistent robots are not required to detour for recharging or battery swapping. We formulate this problem as generalized multiple depots traveling salesman problem (GMDTSP) on a complete graph. An efficient centralized heuristic-based algorithm Multiple Depots Random Select (MDRS) is proposed, which has been proved to have an analytical constant upper bound in the worst case scenario. To provide scalability and robustness, a fully decentralized asynchronous MDRS (dec-MDRS) is proposed, with the analysis of its correctness and efficiency. We also propose a post-processing heuristic (MDRS-IM) to improve the solution quality. We demonstrate the efficiency and effectiveness of our algorithm via benchmark on multiple instances from TSPLIB and GTSPLIB. The simulation results show that the complexity of dec-MDRS grows linearly as the number of robots increases. The simulation also shows that MDRS and MDRS-IM perform significantly faster than the state of the art heuristic solver LKH with only a loss of about 10% of solution quality.

Collision-Aware Task Assignment for Multi-Robot Systems
Wu, Fang;   Varadharajan, Vivek shankar;   Beltrame, Giovanni
We propose a novel formulation of the collision-aware task assignment (CATA) problem and a decentralized auction-based algorithm to solve the problem with optimality bound. Using a collision cone, we predict potential collisions and introduce a binary decision variable into the local reward function for task bidding. We further improve CATA by implementing a receding collision horizon to address the stopping robot scenario, i.e. when robots are confined to their task location and become static obstacles to other moving robots. The auction-based algorithm encourages the robots to bid for tasks with collision mitigation considerations. We validate the improved task assignment solution with both simulation and experimental results, which show significant reduction of overlapping paths as well as deadlocks.

Anytime Multi-Arm Task and Motion Planning for Pick-And-Place of Individual Objects Via Handoffs
Shome, Rahul;   Bekris, Kostas E.
Automation applications are pushing the deployment of many high DoF manipulators in warehouse and manufacturing environments. This has motivated many efforts on optimizing manipulation tasks involving a single arm. Coordinating multiple arms for manipulation, however, introduces additional computational challenges arising from the increased DoFs, as well as the combinatorial increase in the available operations that many manipulators can perform, including handoffs between arms. The focus here is on pick-and-place tasks, which require a sequence of handoffs to be executed, so as to achieve computational efficiency, asymptotic optimality and practical anytime performance. The paper leverages recent advances in multi-robot motion planning for high DoF systems to propose a novel multi-modal extension of the dRRT* algorithm. The key insight is that, instead of naively solving a sequence of motion planning problems, it is computationally advantageous to directly explore the composite space of the integrated multi-arm task and motion planning problem, given input sets of possible pick and handoff configurations. A proof sketch regarding the asymptotic optimality of the method is included in this work given the sampling of additional picks and handoffs over time. The evaluation shows that the approach finds initial solutions fast and improves their quality over time. It also succeeds in finding solutions to harder problem instances relative to alternatives and can scale effectively as the number of robots increases.

Poster Highlights II

Thursday, 22 Aug, 11:30am-11:45am
Session Chair: Rahul Shome

Cooperative Control of Autonomous Floating Modular Structure without Communication
Wang, Wei;  Mateos, Luis;  Wang, Zijian;  Huang, Kuan Wei;  Schwager, Mac;  Ratti, Carlo;  Rus, Daniela
This paper presents a smart modular structure that is able to cooperatively track desired trajectories, orientations, and velocities in a planar water environment without direct communication among modules. A leader in the structure can steer the whole group to these predefined states by applying a relatively small force and torque, whose effect is amplified by the follower robots as they match their forces and torques to the leader’s. We propose a decentralized force and torque controller for each follower which requires only local measurements without communication. Our approach is verified by simulations with 65 robots and experiments with three robotic boats successfully tracking the desired trajectory, orientations, and velocities in a swimming pool.

Towards a Scalable, Self-Reconfigurable Robot with Compliant Modules
Ceron, Steven;  Horowitz, Logan;  Wilson, Nialah Jenae;  Chen, Claire;  Kim, Daniel S;  Petersen, Kirstin Hagelskjaer
We present a planar, self-reconfigurable modular robot with onboard processing, sensing, communication, and passive components for actuation. Modules consist primarily of a flexible printed circuit board rolled in a loop that is populated with all electronic components and 12 simplified electro-permanent magnets for mobility and communication. The flexible PCB eases assembly and introduces compliance for safe interaction with external objects and other modules, as well as configuration of large scale lattice configurations beyond what the manufacturing tolerances allow. This thrust towards scalable robots composed of passively compliant modules is an important step in addressing the grand challenges of mechanical wear and error tolerance in large-scale collectives.

A Soft-Bodied Modular Reconfigurable Robotic System Composed of Interconnected Kilobots
Pratissoli, Federico;  Reina, Andreagiovanni;  Kaszubowski Lopes, Yuri;  Sabattini, Lorenzo;  Gross, Roderich
We describe the Kilobot Soft Robot, a novel soft-bodied robot that is modular and reconfigurable. The Kilobot Soft Robot is realized by inter-connecting a group of miniature mobile modules, based on the commercially available Kilobot, through an elastic material. It moves and deforms fully autonomously. Each module executes a distributed algorithm that exploits only information that is locally obtained using omnidirectional, infrared based signaling. A series of experiments were conducted to validate the algorithm, investigating the ability of the robot to follow a predefined trajectory, to squeeze and extend its shape and to control its motion independently of the number of modules.

A Bio-Inspired Transportation Network for Scalable Swarm Foraging
Lu, Qi;  Moses, Melanie
Scalability is a significant challenge for robot swarms. Generally, larger groups of cooperating robots produce more inter-robot collisions, and in swarm robot foraging, larger search arenas result in larger travel costs. This paper demonstrates a scale-invariant swarm foraging algorithm that ensures that each robot finds and delivers resources to a central collection zone at the same rate, regardless of the size of the swarm or the search area. Dispersed mobile depots aggregate locally foraged resources and transport them to a central place via a hierarchical branching transportation network. This approach is inspired by ubiquitous fractal branching networks such as tree branches and animal cardiovascular networks that deliver resources to cells and determine the scale and pace of life. We demonstrate that biological scaling laws predict how quickly robots forage in simulations of up to thousands of robots searching over thousands of square meters. We then use biological scaling predictions to determine the capacity of depot robots in order to overcome scaling constraints and produce scale-invariant robot swarms. We verify the predictions using ARGoS simulations.

Paper Presentation II: Aerial and marine applications

Thursday, 22 Aug, 11:45am-12:45pm
Session Chair: Giovanni Beltrame

Monitoring Access to User Defined Areas with Swarms of UAVs in Urban Environments
Gupta, Manas;   Lin, Ming C.;   Manocha, Dinesh;   Xu, Huan;   Otte, Michael W.
We present an algorithm that determines where the members of a multi-agent team or swarm should be deployed in order to efficiently monitor access to a user specified region of interest. Our algorithm attempts to minimize the number of agents required to guarantee that any incursion into the region of interest is detected. The algorithm works by analyzing the geometric structure of the environment, and placing agents at advantageous positions in the environment, such as bottlenecks, to create a defensive perimeter of agents alongside physical obstacles (e.g. buildings). We demonstrate the usefulness of the algorithm through experimental simulations in an urban environment, and show how the min-cuts subroutine (used to reduce the number of agents required) can be implemented in a distributed way across the multi-agent team to enable better solutions to be found more quickly.

Mobile Radiation Source Interception by Aerial Robot Swarms
Yadav, Indrajeet;   Tanner, Herbert G.
This paper presents a motion planning approach to track a moving target for the purpose of classifying its radioactivity properties, using a team of ac{mav}. The team falls into an optimal, from a sensing perspective, formation around the target and collectively decide, within seconds, if the target is even weakly radioactive. The ac{mav} formation is optimal in the sense that the vehicles minimize the detection error in their collective decision-making. Intuition from the analytical solution of the basic problem formulation, coupled with numerical simulations, guides the technical approach in which an intractable nonlinear optimization problem is converted into a ac{qp}. The ac{qp} solution then informs a motion planner based on navigation functions, to command the ac{mav} into the desired formation. The complete motion planning and control architecture is tested in simulation.

Multi-Robot Navigation Support System for a Team of Underwater Robots
Quraishi, Anwar Ahmad;   Bahr, Alexander;   Schill, Felix;   Martinoli, Alcherio
The primary and somewhat interrelated challenges affecting deployment of Autonomous Underwater Vehicles (AUVs) are navigation and communication. We have developed miniature, agile, easy to carry and deploy AUVs equipped with a suite of sensors for underwater environmental sensing. In this paper, we propose a support system for multiple AUVs where a group of Autonomous Surface Vehicles (ASVs) coordinate to provide external positioning reference. They transmit an acoustic ranging pulse and then broadcast their position using acoustic communication. Communication errors are detected by using a novel approach where data decoding is coupled with navigation. Our system achieves scalability in the number of AUVs by using one-way travel time for ranging and making the AUVs passive receivers. Further, it allows the ASVs to be repositioned during a mission, so that they can provide positioning aid from a closer range. This is an advantage especially in shallow water environments, where range measurement errors increase significantly with increasing distance. We describe our system in detail and evaluate it with simulations based on real data as well as field experiments.

Trajectory Planning for the Shapeshifting of Autonomous Surface Vessels
Gheneti, Banti;   Park, Shinkyu;   Kelly, Ryan;   Meyers, Drew;   Leoni, Pietro;   Ratti, Carlo;   Rus, Daniela
We present a trajectory planning algorithm for the shapeshifting of reconfigurable modular surface vessels. Each vessel is designed to latch with and unlatch from other vessels, which we aim to use to create dynamic infrastructure, such as on-demand bridges and temporary market squares, in canal environments. Our algorithm generates smooth and collision-free trajectories that the vessels can track to reconfigure their connections. We formulate the trajectory planning problem as Mixed Integer Quadratic Programming (MIQP) with a B-spline representation. We conceive a physical platform 1:4 prototypes of the reconfigurable modular vessels and, through swimming pool experiments, show the efficacy of our trajectory planning algorithm for the shapeshifting of the vessels.

Poster Highlights III

Thursday, 22 Aug, 2pm-2:15pm
Session Chair: Alberto Quattrini Li

Decentralized Dynamic Task Allocation in Swarm Robotic Systems for Disaster Response
Ghassemi, Payam;  DePauw, David;  Chowdhury, Souma
Multiple robotic systems, working together, can provide important solutions to different real-world applications (e.g., disaster response), among which task allocation problems feature prominently. Very few existing decentralized multi-robotic task allocation (MRTA) methods simultaneously offer the following capabilities: consideration of task deadlines, consideration of robot range and task completion capacity limitations, and allowing asynchronous decision-making under dynamic task spaces. To provision these capabilities, this paper presents a computationally efficient algorithm that involves novel construction and matching of bipartite graphs. Its performance is tested on a multi-UAV flood response application.

Multirobot Simultaneous Path Planning and Task Assignment on Graphs with Stochastic Costs
Yang, Fan;  Chakraborty, Nilanjan
We present a novel algorithm for problems involving both task assignment and stochastic path planning on a road map. In this problem, the initially unassigned robots and tasks are located at known position in a road map. We want to assign a unique task to each robot and compute path for robot going to the task location. Given the means and variances of travel cost, our goal is to develop algorithms that guarantee that for each robot, the total travel cost is below a minimum value in any realization of the stochastic travel costs with high probability. We prove that the solution can be obtained by solving chance-constrained shortest path problems for all robot-task pairs and a linear bottleneck assignment problem in which the cost of an assignment is equal to the optimal objective value of the former problem. We propose algorithms for solving chance-constrained shortest path problem either optimally or approximately by solving a number of deterministic shortest path problems that minimize some linear combination of means and variances of edge costs. We present simulation results on randomly generated networks and data to demonstrate that our algorithm is scalable with the number of robots (or tasks) and the size of the network.

Self-Reactive Planning of Multi-Robots with Dynamic Task Assignments
Yang, Qin;  Luo, Zhiwei;  Song, Wenzhan;  Parasuraman, Ramviyas
Multi-Robot Systems and Swarms are intelligent systems in which a large number of agents are coordinated in a distributed and decentralized way. Each robot may have homogeneous or heterogeneous capabilities and can be programmed with several fundamental control laws adapting to the environment. Through different kinds of relationships built using the communication protocols, they present various behaviors based on the shared information and current state. To adapt to dynamic environments effectively, and maximize the utility of the group, robots need to cooperate, share their local information, and make a suitable plan according to the specific scenario. In this paper, we formalize the problem of multi-robots fulfilling dynamic tasks using state transitions represented through a Behavior Tree. We design a framework with corresponding distributed algorithms for communications between robots and negotiation and agreement protocols through a novel priority mechanism. Finally, we evaluate our framework through simulation experiments.

Multi-Agent Pursuit-Evasion under Uncertainties with Redundant Robot Assignments
Zhang, Leiming;  Prorok, Amanda;  Bhattacharya, Subhrajit
We consider a pursuit-evasion problem with a heterogeneous team of multiple pursuers and multiple evaders. The pursuers (robots), using only noisy on-board sensors, can make a probabilistic estimation of positions of multiple moving evaders based on sensor measurements of signals emitted by the evaders. The evaders being aware of the environment and the position of all pursuers follow a strategy to actively avoid being intercepted. We model the evaders’ motion as a time-varying Markov process, and along with stochastic measurements, the pursuers use Markov Localization to update the probability distribution of the evaders. A search-based motion planning strategy is developed that intrinsically takes the probability distribution of the evaders into account. Pursuers are assigned using an assignment algorithm that takes redundancy into account, such that the estimated net time to capture the evaders is minimized. Our experimental evaluation shows that the redundant assignment algorithm performs better than an alternative nearest-neighbor based assignment algorithm.

Scheduling Spare Drones for Persistent Task Performance with Several Replacement Stations
Hartuv, Erez;  Agmon, Noa;  Kraus, Sarit
This paper considers the problem of enabling persistent execution of a multi-drone task under energy limitations when several battery replacement stations (homes) are available. The drones are given a set of locations and their task is to ensure that at least one drone will be present, for example for monitoring, over each location at any given time. Because of energy limitations, drones must be replaced from time to time, and fly back to one of a given fixed set of homes where their batteries can be replaced. Our goals are to identify the minimum number of spare drones needed to accomplish the task while no drone battery drains, and to provide a drone replacement strategy.

Keynote by Bryan Low

Thursday, 22 Aug, 2:15pm-3:00pm
Introduced by Dylan Shell
Collective Learning and Model Fusion in Large Multi-Agent Systems

Collective learning and model fusion are emerging areas that aim to pave the way forward to broaden the deployability of AI/ML in multi-agent systems. This is motivated from a limitation of existing systems which often deploy AI in narrow, offline and single-agent contexts, thus resulting in in-house/local solutions using local/in-house resources that fail to utilize the collective computation and communication capability of the entire system to support its ever-growing scale. In this talk, I will define collective learning and its key desiderata of (a) exploiting the collective resources for computation via on-demand model fusion; (b) representing local knowledge efficiently for peer-to-peer communication; and (c) developing a decentralized knowledge propagation algorithm to ensure consensus across the entire growing crowd of agents. I will also discuss the challenge of deploying collective learning in information-sensitive domains where independent agents are further constrained with the heterogeneity and privacy of their model architecture, which increases the difficulty in achieving the above desiderata. To this end, I will present our preliminary development of a set of principled solution techniques to make the first step towards achieving the above goals.

Paper Presentation III: Swarms

Thursday, 22 Aug, 4pm-5:15pm
Session Chair: Philip Dames

Robots That Sync and Swarm: A Proof of Concept in ROS 2
Barciś, Agata;   Barciś, Michał   Bettstetter, Christian
A unified mathematical model for synchronisation and swarming has recently been proposed. Each system entity, called a "swarmalator”, coordinates its internal phase and location with the other entities in a way that these two attributes are mutually coupled. This paper realises and studies, for the first time, the concept of swarmalators in a technical system. We adapt and extend the original model for its use with mobile robots and implement it in the Robot Operating System 2 (ROS 2). Simulations and experiments with small robots demonstrate the feasibility of the model and show its potential to be applied to real-world systems. All types of space-time patterns achieved in theory can be reproduced in practice. Applications can be found in monitoring, exploration, entertainment and art, among other domains.

A Minimalistic Approach to Segregation in Robot Swarms
Mitrano, Peter;   Burklund, Jordan;   Giancola, Michael;   Pinciroli, Carlo
We present a decentralized algorithm to achieve segregation into an arbitrary number of groups with swarms of autonomous robots. The distinguishing feature of our approach is in the minimalistic assumptions on which it is based. Specifically, we assume that (i) Each robot is equipped with a ternary sensor capable of detecting the presence of a single nearby robot, and, if that robot is present, whether or not it belongs to the same group as the sensing robot;   (ii) The robots move according to a differential drive model;   and (iii) The structure of the control system is purely reactive, and it maps directly the sensor readings to the wheel speeds with a simple `if’ statement. We present a thorough analysis of the parameter space that enables this behavior to emerge, along with an analysis that explains the emergent behavior. Finally, we study the effect on segregation performance of some non-ideal aspects in the robot morphology.

Distributed Algorithm for Selecting Leaders for Supervisory Robotic Swarm Control
Lewkowicz, Michal;   Agarwal, Rohil;   Chakraborty, Nilanjan
In this paper, we present a distributed algorithm for selecting multiple leaders in a swarm that can be used for supervisory control of the swarm system. The usage of optimally placed leaders within swarms can minimize the communication requirements for information dissemination across a robotic network. We formulate the leader selection problem as a combinatorial optimization problem and provide a novel characterization of the optimal solution based on the notion of Voronoi decomposition of a graph. Based on this characterization, we present an incremental, distributed algorithm to compute the leader set. We present simulation results to show that our approach results in optimal performance.

The Robotic Swarm Contamination Problem
Avrahami, Sapir;   Agmon, Noa
The problem of reaching a consensus in a swarm of robots has been widely examined in the literature. In this paper we introduce the contamination problem, a new version of the consensus problem in which the robots’ influence to adopt a certain value towards a consensus is based on both external factors (its surroundings) and its internal state. The problem’s innovation is in the combination of two aspects: first, the existence of adversarial forces that may intentionally act to divert the robots from a desired consensus, and an asymmetry of state transition, that is, a robot may have a stronger tendency to change from one state to another, compared to the other way around. We provide a theoretical analysis of formation creation as a mean to guarantee consensus, and an empirical evaluation – both in simulation and in real robots – demonstrating the influence of different robot behaviors and internal factors on the convergence time and value.

Communication through Motion: Legibility of Multi-Robot Systems
Capelli, Beatrice;   Secchi, Cristian;   Sabattini, Lorenzo
The interaction between a user and a multi-robot system in a shared environment is a relatively uncharted topic. But, as these types of systems will increase in the future years, an efficient way of communication is necessary. To this aim, it is interesting to discover if a multi-robot system can communicate its intentions exploiting only some motion-variables, which are characteristics of the motion of the robots. This study is about the legibility of a multi-robot system: in particular, we focus on the influence of these motion-variables on the legibility of more than one group of robots that move in a shared environment with the user. These motion-variables are: trajectory, dispersion and stiffness. They are generally used to define the motion of a group of mobile robots. Trajectory and dispersion were found relevant for the correctness of the communication between the user and the multi-robot system, while stiffness was found relevant for the rapidity of communication. The analysis of the influence of the motion-variables was carried out with an ANOVA (analysis of variance) based on a series of data coming from an experimental campaign conducted in a virtual reality set-up.

Keynote by Dan Halperin

Thursday, 22 Aug, 5:15pm-6:00pm
introduced by Kostas Bekris
Multi-Robot Motion Planning: The Easy, the Hard and the Uncharted

There are multi-robot motion planning (MRMP) problems involving dozens of robots, which can be speedily solved, while other are practically unsolvable. What makes an MRMP problem easy or hard? The first part of the talk will describe our quest to resolve this issue, and some progress we have made in the context of unlabeled MRMP.
The second part of the talk will review recent algorithms that we have developed for various types of MRMP problems in tight obstacle-cluttered environments: from sampling-based methods tailored to the coordination of a few complex robots, to complete and exact solutions for hundreds of simply shaped robots. The talk will conclude with surprisingly open problems for the coordinated motion of two robots.

Click here to go back to the top of the page.
Click here to see the Friday program on PaperCept.

Friday, 23 Aug

Keynote by Katia Sycara

Friday, 23 Aug, 9:00am-9:45am
Introduced by Chris Amato
Trustworthy Human Teaming with Multi-Robot Systems

As robots become part of the fabric of human life, devising computational models for robot interactions with other robots as well as interaction with humans is rapidly becoming an increasingly important research area. To realize this future, robots must be able to autonomously coordinate and also develop capabilities of teaming with humans. Explicit centralized control becomes increasingly impractical as robots grow in number and capabilities. On the other hand, emergent group behaviors that may result from decentralized control are particularly ill suited for humans to predict or interact with. In particular, emergent swarm performance presents challenges for human trust which may result in maladaptive human-autonomy trust calibration. This could be problematic since trust has been shown to be an important component of human-autonomy teams. In this talk, I will present our recent work in swarm coordination, human-swarm teaming, methods for assessing human trust, and optimal timing for switching between swarm behaviors

Poster Highlights IV

Friday, 23 Aug, 9:45am-10am
Session Chair: Kiril Solovey

Computational and Structural Advantages of Pairwise Flocking
Nagy, Geoff;  Thornton, Alex;  Ling, Hangjian;  McIvor, Guillam;  Ouellette, Nicholas;  Vaughan, Richard
Biologists have shown that certain species of birds flock in groups at least partially composed of pairs, believed to be stable mate pairs. In this paper we consider synthetic flocks containing pairs, and show that this structure can be used to reduce paired agents’ tracking loads by approximately half, while also providing a more stable flock despite the fewer tracking interactions. We suggest this could be useful for constructing robust, real-world flocking systems.

Non-Uniform Policies for Multi-Robot Asymmetric Perimeter Patrol in Adversarial Domains
Oshrat, Yaniv;  Agmon, Noa;  Kraus, Sarit
In this work we examine the {em Closed Perimeter Patrol in Adversarial Domain} problem, in which robots travel along a closed perimeter and the adversary is aware of the robots’ patrol policy. Unlike former works which focused on symmetric tracks, we explore the more realistic asymmetric track scenarios and show that in non-deterministic patrol schemes, non-uniform probability models are better than uniform ones.

Decentralised self-organising maps for the online orienteering problem with neighbourhoods
Best, Graeme;  Hollinger, Geoffrey
This extended abstract presents a new coordination algorithm for decentralised multi-robot information gathering. We consider planning for an online variant of the multi-agent orienteering problem with neighbourhoods. This formulation closely aligns with a number of important tasks in robotics, including inspection, surveillance, and reconnaissance. We propose a decentralised variant of the self-organising map (SOM) learning procedure, named Dec-SOM, which efficiently finds plans for a team of robots in a non-myopic manner. Decentralisation is achieved by performing a distributed allocation scheme jointly with the SOM adaptations. Preliminary simulation results indicate that Dec-SOM outperforms baseline methods and is a viable solution for decentralised, online, and non-myopic information gathering.

Towards an Online Approach for Knowledge Communication Planning
Tsiogkas, Nikolaos;  Lane, David
Explicit coordination among robot usually requires a method to exchange information regarding their intentions and their knowledge of the world. Most of the research found in the multi-robot coordination literature uses ad-hoc approaches for communication. This work is an initial attempt to tackle the communication problem in a generic way. It tries to answer the question of when to communicate, what to communicate and with whom to communicate. It proposes the use of dynamic action sets combined with a Determinized Sparse Partially Observable Tree structure that can address the communication decision problem in an online fashion. Compared with a state of the art approach used for sharing knowledge among robotic teams it shows a constantly better performance.

Memory-Based Multiagent One-Shot Learning
Khadka, Shauharda;  Yates, Connor;  Tumer, Kagan
Learning correct behavior from one example (one-shot learning) is particularly difficult in multiagent systems where the pertinent information is potentially distributed across agents, and the emergent behavior of the system is dependent on inter-agent interactions. This paper introduces Distributed Modular Memory Units (DMMU), a distributed multiagent learning framework. DMMU uses a shared external memory to enable one-shot adaptive learning in multiagent systems. The external memory is accessed by agents acting independently and in parallel. Agents operate by processing their individual states and deciding when to interact with the shared external memory unit. This allows agents to identify, retain, and propagate information among the team. Subsequently, DMMU can rapidly assimilate task features from a group of distributed agents and successfully use the information to accomplish distributed one-shot learning. We compare the performance of the DMMU framework on a simulated cybersecurity task with traditional feed-forward ensembles, long short-term memory based agents, and a centralized learner. Results demonstrate that DMMU shows at least 100% performance improvement over the other methods, and exhibits distributed one-shot learning to effectively solve this complex task.

Paper Presentation IV: Coverage control and collision avoidance

Friday, 23 Aug, 10am-11am
Session Chair: Nilanjan Chakraborty

Voronoi-Based Coverage Control with Connectivity Maintenance for Robotic Sensor Networks
Luo, Wenhao;   Sycara, Katia
In this paper, we consider the problem of Voronoi-based coverage control for multi-robot sensor networks with connectivity constraints, where a team of mobile robots spread out over the workspace in order to optimize the overall coverage performance while preserving connectivity. In particular, we proposed a connectivity-aware coverage control approach to simultaneously optimize the coverage performance and ensure robotic network connectivity. Unlike most of existing works on connectivity control that have no guarantee on the primary task, our algorithm could maintain minimum connectivity that introduce minimal revision to the primary coverage controller due to invoked connectivity constraints. This ensures the robotic network stay connected at all time but also in an optimal way to provide highest freedom for achieving primary coverage task. Moreover, we prove the convergence of the proposed controller that guarantees continuously improved coverage performance in presence of connectivity constraints. Simulation results are given to demonstrate the effectiveness of our approach.

Decentralized Minimum-Energy Coverage Control for Time-Varying Density Functions
Santos, María;   Mayya, Siddharth;   Notomista, Gennaro;   Egerstedt, Magnus
This paper introduces a minimum-energy approach to the problem of time-varying coverage control. The coverage objective, encoded by a locational cost, is reformulated as a constrained optimization problem that can be solved in a decentralized fashion. This allows the robots to achieve a centroidal Voronoi tessellation by running a decentralized controller even in case of a time-varying density function. We demonstrate that this approach makes no assumptions on the rate of change of the density function and performs the computations in an approximation-free manner. The performance of the algorithm is evaluated in simulation as well as on a team of mobile robots.

B-UAVC: Buffered Uncertainty-Aware Voronoi Cells for Probabilistic Multi-Robot Collision Avoidance
Zhu, Hai;   Alonso-Mora, Javier
This paper presents B-UAVC, a distributed collision avoidance method for multi-robot systems that accounts for uncertainties in robot localization. In particular, Buffered Uncertainty-Aware Voronoi Cells (B-UAVC) are employed to compute regions where the robots can safely navigate. By computing a set of chance constraints, which guarantee that the robot remains within its B-UAVC, the method can be applied to non-holonomic robots. A local trajectory for each robot is then computed by introducing these chance constraints in a receding horizon model predictive controller. The method guarantees, under the assumption of normally distributed position uncertainty, that the collision probability between the robots remains below a specified threshold. We evaluate the proposed method with a team of quadrotors in simulations and in real experiments.

Distributed Collision Avoidance of Multiple Robots with Probabilistic Buffered Voronoi Cells
Wang, Mingyu;   Schwager, Mac
This paper introduces Probabilistic Buffered Voronoi Cell (PBVC) collision avoidance for multiple robots using noisy on-board sensing. The work builds upon the previously proposed Buffered Voronoi Cell (BVC) approach. We introduce a probabilistic formulation to construct a family of BVCs with specified safety levels, which take into account uncertainty in sensor measurements among the robots. The safety level of a PBVC represents the probability that the area is contained inside the robot’s true (but unknown) BVC. The PBVC provides a set of probabilistically safe positions for the robot to navigate to. Each agent chooses its next position constrained within the PBVC of a desired safety level, while minimizing the deviation from its reference trajectory. We prove a conservative bound on the probability of collision given this reciprocal strategy. We also validate through simulations that the proposed approach achieves safe navigation among multiple robots in challenging scenarios, and provides a significantly lower risk of collision than either the Reciprocal Velocity Obstacles (RVO) method or the Buffered Voronoi Cell (BVC) method when the robots use noisy relative measurements. We also show in experiments with small-scale robotic cars that the algorithm is fast, effective and is useful in real applications.

Poster Highlights V

Friday, 23 Aug, 11:30am-11:45am
Session Chair: Alberto Quattrini Li

Decentralized Nonlinear Model Predictive Control for 3D Formation of Multirotor Micro Aerial Vehicles with Relative Sensing and Estimation
Erunsal, Izzet Kagan;  Martinoli, Alcherio;  Ventura, Rodrigo
The coordination and cooperation strategies of Micro Aerial Vehicles (MAVs) become increasingly popular to carry out high-level, complex tasks such as object transportation, patrolling, inspection robustly and with high performance. One of the crucial component of these strategies is formation control. Most formation control approaches assume absolute localization and/or leverage communication between robots. In contrast, the main focus of this paper is three-dimensional decentralized formation control of multi-rotor MAVs by using exclusively relative sensing and estimation, eliminating any explicit communication between robots. The proposed methodology is Decentralized Nonlinear Model Predictive Control (D-NMPC) approach, to aim optimality while satisfying constraints and scalability, in a leader-follower scheme in conjunction with Pose-Graph Moving Horizon Estimator (PG-MHE), to estimate neighbour state while filtering noise effectively. After introducing a realistic six degrees of freedom (DOF) mathematical model of MAVs, the problem is formulated based on the local coordinate frames of robots. This type of formulation makes the formation independent of the full knowledge of global or common reference frames. Then, PG-MHE is presented to filter out and estimate the required states by D-NMPC. The next step is to propose a D-NMPC strategy, construct an Optimal Control Problem (OCP) and solve it by modified Real-time Iteration (RTI) scheme. A comprehensive simulative scenario consists of three quadrotors is designed to observe and validate the convergence of the method. The initial results of this study show that satisfactory performance and convergence are achieved under noise in all local sensors and in cases where the dynamics of the formation quickly changes.

Multi-Robot Control Using Coverage Over Time-Varying Domains
Xu, Xiaotian;  Diaz-Mercado, Yancy
An approach is proposed to influence the motion of a group of robots by using time-varying domains and multi-agent coverage control. We synthesize controllers for the individual agents by specifying the desired spread and motion of the multi-robot team as a whole. The result is an approach that is agnostic to the size of the multi-robot team and that does not require assignments of roles, e.g., leaders and followers. The modified coverage optimization problem is stated to accommodate time-varying domains, and a control strategy is proposed to solve the problem with exponential convergence guarantees. Analytical expressions for the surface and line integral terms in the proposed coverage control law can be provided for the uniform coverage density case. A robot implementation of the control strategy on a multi-robot team is used to validate the results.

Optimal Policies for Routing Multi Sensor-Effector Combined Autonomous Devices
Oh, Yura;  Powell, Warren
We use Monte Carlo tree search to create policiesthat simultaneously performs active learning about the con-centration of contaminants using a Bayesian prior, while alsoperforming mitigation. The policy is asymptotically optimal forone device. We then propose and compare different informationsharing protocols to coordinate a fleet.

Optimized Motion Strategy for Active Target Localization of Mobile Robots with Time-Varying Connectivity
Zhang, Liang;  Zhang, Zexu;  Siegwart, Roland;  Chung, Jen Jen
This paper addresses optimal trajectory generation for active target positioning using a collective localization scheme under a time-varying observation topology. We show that a team of assisting robots using the optimal trajectories can improve the localization accuracy of leader robots whose commands are assigned by high-level tasks. We apply the standard centralized extended Kalman filter to estimate all robot positions by using distance-only relative measurements. In this work, we also explicitly consider the limits on the maximum ranging distance within which the robots are able to make pairwise measurements. The trace of the covariance sub-matrix corresponding to the leader robot’s position estimate is selected as the optimization criterion. Simulation results are presented that demonstrate the applicability of this method and provide insights into the difficulties in optimizing this problem.

Distributed Multi-Target Search and Tracking Using a Coordinated Team of Ground and Aerial Robots
Chen, Jun;   Dames, Philip;  
The authors recently developed a distributed algorithm to enable a team of homogeneous robots to search for and track an unknown and time-varying number of dynamic targets. This algorithm combined a distributed version of the PHD filter (for multi-target tracking) with Lloyd’s algorithm to drive the motion of the robots. In this paper we extend this previous work to allow a heterogeneous team of ground and aerial robots to perform the search and tracking tasks in a coordinated manner. Both types of robots are equipped with sensors that have a finite field of view and which may receive both false positive and false negative detections. The aerial robots may vary the size of their sensor field of view (FoV) by changing elevation. This increase in the FoV coincides with a decrease in the accuracy and reliability of the sensor. The ground robots maintain the target tracking information while the aerial robots provide additional sensor coverage. We develop two new distributed algorithms to provide filter updates and to make control decisions in this heterogeneous team. Both algorithms only require robots to communicate with nearby robots and use minimal bandwidth. We demonstrate the efficacy of our approach through a series of simulated experiments which show that the heterogeneous teams are able to achieve more accurate tracking in less time than our previous work.

Paper Presentation V: Exploration, mapping, and search

Friday, 23 Aug, 11:45am-12:45pm
Session Chair: Carlo Pinciroli

Informative Path Planning with Local Penalization for Decentralized and Asynchronous Swarm Robotic Search
Ghassemi, Payam;   Chowdhury, Souma
Decentralized swarm robotic solutions to searching for targets that emit a spatially varying signal promise task parallelism, time efficiency, and fault tolerance. It is, however, challenging for swarm algorithms to offer scalability and efficiency, while preserving mathematical insights into the exhibited behavior. A new decentralized search method (called {decBayes}), founded on batch Bayesian Optimization (BO) principles, is presented here to address these challenges. Unlike swarm heuristics approaches, {decBayes} decouples the knowledge generation and task planning process, thus preserving insights into the emergent behavior. Key contributions lie in: 1) modeling knowledge extraction over trajectories, unlike in BO;   2) time-adaptively balancing exploration/exploitation and using an efficient local penalization approach to account for potential interactions among different robots’ planned samples;   and 3) presenting an asynchronous implementation of the algorithm. This algorithm is tested on case studies with bimodal and highly multimodal signal distributions. Up to 76 times better efficiency is demonstrated compared to an exhaustive search baseline. The benefits of exploitation/exploration balancing, asynchronous planning, and local penalization, and scalability with swarm size, are also demonstrated.

Decentralized Multi-Floor Exploration by a Swarm of Miniature Robots Teaming with Wall-Climbing Units
Leong Kit, Jabez;   Dharmawan, Audelia Gumarus;   Mateo, David;   Foong, Shaohui;   Soh, Gim Song;   Bouffanais, Roland;   Wood, Kristin
In this paper, we consider the problem of collectively exploring unknown and dynamic environments with a decentralized heterogeneous multi-robot system consisting of multiple units of two variants of a miniature robot. The first variant—a wheeled ground unit—is at the core of a swarm of floor-mapping robots exhibiting scalability, robustness and flexibility. These properties are systematically tested and quantitatively evaluated in unstructured and dynamic environments, in the absence of any supporting infrastructure. The results of repeated sets of experiments show a consistent performance for all three features, as well as the possibility to inject units into the system while it is operating. Several units of the second variant—a wheg-based wall-climbing unit—are used to support the swarm of mapping robots when simultaneously exploring multiple floors by expanding the distributed communication channel necessary for the coordinated behavior among platforms. Although the occupancy-grid maps obtained can be large, they are fully distributed. Not a single robotic unit possesses the overall map, which is not required by our cooperative path-planning strategy.

Map Merging of Oriented Topological Semantic Maps
Susa Rincon, Jose Luis;   Carpin, Stefano
In this paper we propose a solution for the problem of merging together partial spatial models relying on our recently introduced Oriented Topological Semantic Maps (OTSM). This problem arises when a group of robots cooperatively explores an environment and each robot independently builds a partial map that needs to be combined into a full map. Our methodology is inspired by the Inverse Warrington’s Object Recognition Model (IWORM), a cognitive model hypothesizing two post-sensory categorical stages working together for object recognition. Inspired by IWORM, our approach uses a two step approach to compare places in different maps and match them together depending of their mutual resemblance. The method is complemented by a scoring system to measure the likelihood that two vertices in different OTSMs correspond to the same vertex, despite possible errors in labeling, orientation, or topological structure. Our methodology is validated in simulation, thus allowing to perform various experiments with carefully controlled error sources.

Dirichlet-Multinomial Counterfactual Rewards for Heterogeneous Multiagent Systems
Dixit, Gaurav;   Zerbel, Nicholas;   Tumer, Kagan
Multi-robot teams have been shown to be effective in accomplishing complex tasks which require tight coordination among team members. In homogeneous systems, recent work has demonstrated that "stepping stone"   rewards are an effective way to provide agents with feedback on potentially valuable actions even when the agent-to-agent coupling requirements of an objective are not satisfied. In this work, we propose a new mechanism for inferring hypothetical partners in tightly-coupled, heterogeneous systems called Dirichlet-Multinomial Counterfactual Selection (DMCS). Using DMCS, we show that agents can learn to infer appropriate counterfactual partners to receive more informative stepping stone rewards by testing in a modified multi-rover exploration problem. We also show that DMCS outperforms a random partner selection baseline by over 40%, and we demonstrate how domain knowledge can be used to induce a prior to guide the agent learning process. Finally, we show that DMCS maintains near optimal performance for up to 15 distinct rover types while the performance of the baseline degrades.

Poster Highlights VI

Friday, 23 Aug, 2pm-2:15pm
Session Chair: Shuai Han

An Approximation Algorithm for Distributed Resilient Submodular Maximization
Zhou, Lifeng;  Tokekar, Pratap
We study a distributed coordination problem in which a group of robots collaborates to maximize a submodular objective function. We consider an adversarial scenario where a subset of the robots can be attacked and cannot contribute to the objective function. We do not know which robots will be attacked, but know an upper bound of how many can be attacked. We study the distributed version of the problem, where each robot must choose its own actions only by communicating with its neighbors. We present a distributed resilient submodular maximization algorithm that guarantees performance within a constant factor of the optimal. Our analysis uses the curvature of submodular set functions. We show that the algorithm is scalable, runs in polynomial time, and is faster than its centralized version. We demonstrate the efficacy of our algorithm through simulations of a multi-robot target tracking scenario.

Data-Efficient Decentralized Place Recognition with 3D Constellations of Objects
Ramtoula, Benjamin;  de Azambuja, Ricardo;  Beltrame, Giovanni
Using multiple robots for exploring and mapping environments can provide improved robustness and performance, but it can be difficult to implement. In particular, limited communication bandwidth is a considerable constraint when a robot needs to determine if it has visited a location that was previously explored by another robot, as it requires for robots to share descriptions of places they have visited. One way to compress this description is to use constellations, groups of 3D points that correspond to the estimate of a set of relative object positions. Constellations maintain the same pattern from different viewpoints and can be robust to illumination changes or dynamic elements. We present a method to extract from these constellations a compact spatial semantic descriptor of the objects in a scene. We use this representation in a 2-step decentralized loop closure verification: first, we distribute the compact semantic descriptors to determine which other robots might have seen scenes with similar objects;  then we query matching robots with the full constellation to validate the match using geometric information. The proposed method requires less memory, is more interpretable than global image descriptors, and could be useful for other tasks and interactions with the environment. We validate our system’s performance on a TUM RGB-D SLAM sequence and show its benefits in terms of bandwidth requirements.

Fitness Critics for Multiagent Learning
Rockefeller, Golden;  Mannion, Patrick;  Tumer, Kagan
This paper presents fitness critics as noise-reduced fitness evaluation functions for cooperative coevolution of multiagent teams. Noise in the evaluation function can have a negative impact on evolutionary learning, especially in multiagent learning. The evaluation function for an individual agent can be noisy as each agent in the team changes their policies independently. Existing noise-reduction methods are not suitable for efficient evolutionary algorithms. The noise-reduction provided by actor-critic methods cannot be directly applied to the evolutionary algorithms when learning policies in episodic problems as the critic provides stepwise evaluations but evolutionary algorithms require episodic fitness evaluations. Fitness critics can provide noise-reduced episodic fitness evaluations by aggregating the multiple stepwise evaluations of an intermediate critic. With demonstrations on the tightly coupled multi-rover domain, this paper shows that teams trained with fitness critics achieve comparable or increased team performance scores compared to teams trained with other noise-reduction methods.

Traffic Management Strategies for Multi-Robotic Rigid Payload Transport Systems
Sirineni, Yahnit;  Verma, Pulkit;  Karlapalem, Kamalakar
In this work, we address traffic management of multiple payload transport systems (PTS) comprising of non-holonomic robots. Each PTS is a loosely coupled rigid robot formation carrying a payload. We ensure each PTS traverses its desired trajectory while avoiding collisions with other PTS and static obstacles in various kinds of complex environments. Each PTS has one leader and multiple followers and the followers maintain a desired distance and angle w.r.t the leader using a decentralized leader-follower control architecture. We showcase through simulations that our strategies help manage traffic for a large number of PTS moving around.

Transportation of Deformable Payload through Static and Dynamic Obstacles Using Loosely Coupled Nonholonomic Robots
Chand, Subhasis;  Verma, Pulkit;  Tallamraju, Rahul;  Karlapalem, Kamalakar
This paper presents our solution for transportation of deformable payloads using a formation of non-holonomic mobile robots while avoiding static and dynamic obstacles. We assume the center of the formation to be a virtual robot which also plays the role of the leader of the formation. Rapidly-exploring Random Tree star (RRT*) algorithm is used to determine a static obstacle free trajectory for the virtual leader. In addition to that, a custom path planning algorithm is incorporated for the follower robots. A decentralized leader-follower formation control is applied for the robots to traverse their respective trajectories without breaking the formation while following the constraints of the deformable payload. The virtual leader moves forward or backward on the initially generated path with variable velocity depending on the proximity of dynamic obstacles. All the other robots follow it so that the formation is able to avoid the dynamic obstacles while still remaining on the path leading towards the destination.

Paper Presentation VI: Formation control and reconfiguration

Friday, 23 Aug, 2:15pm-3:15pm
Session Chair: Noa Agmon

Passivity-Based Decentralized Control of Multi-Robot Systems with Delays Using Control Barrier Functions
Notomista, Gennaro;   Cai, Xiaoyi;   Yamauchi, Junya;   Egerstedt, Magnus
In this paper, we present a solution to the problem of coordinating multiple robots across a communication channel that experiences delays. The proposed approach leverages control barrier functions in order to ensure that the multi-robot system remains dissipative. This is achieved by encoding the dissipativity-preserving condition as a set invariance constraint. This constraint is then included in an optimization problem, whose objective is that of modifying, in a minimally invasive fashion, the nominal input to the robots. The formulated optimization problem is decentralized in the sense that, in order to be solved, it does not require the individual robots to have access to global information. Moreover, owing to its convexity, each robot can solve it using fast and efficient algorithms. The effectiveness of the proposed control framework is demonstrated through the implementation of a formation control algorithm in presence of delays on a team of mobile robots.

A Reinforcement Learning Approach to Multi-Robot Planar Construction
Strickland, Caroline;   Churchill, David;   Vardy, Andrew
We consider the problem of shape formation in a decentralized swarm of robots trained with reinforcement learning. Shapes are formed from ambient objects which are pushed into a desired pattern. The shape is specified using a projected scalar field that the robots can locally sample. This scalar field plays a similar role to the pheromone gradients used by social insects such as ants and termites to guide the construction of their sophisticated nests. The overall approach is inspired by our previously developed orbital construction algorithm. In this paper, we use reinforcement learning to automatically learn policies that accomplish shape formation without the need for hand-coding algorithmic solutions for each desired shape. The particular research questions addressed in this paper are as follows: (1) The performance of learned policies versus the original hard-coded orbital construction algorithm;   (2) The performance of the system on more shapes than were considered for the original algorithm. (3) The impact of the number of robots used in training and then subsequently in testing;   We provide experimental results using a custom two dimensional physics simulator of an environment containing circular robots and objects.

Decentralized Gathering of Stochastic, Oblivious Agents on a Grid: A Case Study with 3D M-Blocks
Ozdemir, Anil;   Romanishin, John;   Gross, Roderich;   Rus, Daniela
We propose stochastic control policies for gathering a group of embodied agents in a two-dimensional square tile environment. The policies are fully decentralized and can be executed on anonymous, oblivious agents with chirality, but no sense of orientation. The agents require only 4 ternary digits of information. We prove that a group of agents, irrespective of initial positions, will almost surely reach a Pareto optimal configuration in finite time. For one of the control policies, computer simulations show that groups of up to 20 agents consistently reach Pareto optimal configurations, whereas groups of 1000 agents, given the same amount of time, improve the compactness of their configurations on average by 89.20%. The policy also copes well with sensory noise up to a level of 50%. We also present an experimental validation using 6 physical 3D M-Block modules, demonstrating the feasibility of the stochastic control approach in practice.

Comparative Analysis of Sensors in Rigid and Deformable Modular Robots for Shape Estimation
Ceron, Steven;   Wilson, Nialah Jenae;   Horowitz, Logan;   Petersen, Kirstin Hagelskjaer
Modular self-reconfigurable robots have the ability to change their connection topology to form large networks of distributed sensors and actuators, for example to reason about gradients in the collective or the geometry of their surroundings. Past work has focused on distributed algorithms to compute the intrinsic shape of the robot. Here, we instead focus on distributed algorithms that permit planar modular robots to compute the shape of external objects in 2D. Grounded in characterization of physical sensors, we simulate shape estimation by modular robots for increasing object size and complexity. Specifically, we compare the use of connectivity and IR sensors which are commonly available on modules, and demonstrate what may be gained from deformable modules with integrated strain sensors. The algorithms we present are dependent only on motion constraints and may generalize to a wide set of modular platforms.

Keynote by Gaurav Sukhatme

Friday, 23 Aug, 4:15pm-5:00pm
Introduced by Jingjin Yu
Resilience by Reconfiguration: Exploiting Heterogeneity in Robot Teams

In this talk I will discuss resilience in multi-robot systems. In the first part of the talk I will attempt to clarify and disambiguate related ideas in multi-robot systems including but not limited to robustness, recovery, and resilience. In the second part of the talk (joint work with Ragesh Ramachandran and James Preiss), I will describe recent work on a method to maintain high resource availability in a networked heterogeneous multi-robot system subject to resource failures. I will describe a model in which sensing and computational resources are available on robots that are engaged in a joint task. When a resource on a particular robot becomes unavailable (e.g., a sensor ceases to function), our method automatically reconfigures the system so that the robot continues to have access to this resource by communicating with other robots. Our method can compute a communication topology, spatial formation, and formation change motion planning in a few seconds. I will show the results of our method operating in simulation and preliminary physical experiments with a small team of seven quadrotors.

Click here to go back to the top of the page.
Click here to go to the program overview page.