Theory, Practice and Applications
Multi-robot collaborative systems are an integral part of any large scale operation. Even more the heterogeneity of the team while opening up great opportunities - different viewpoints, traversability constraints, battery life extension, at the same time creates a new dimensionality of difficulty. Increasing interest in using heterogeneous systems has resulted in numerous works in the literature, nevertheless these solutions tend to divide and conquer - divide tasks for each type of robot and solve for a homogenous system. Moreover, we see very few real world demonstrations in academia and the industry of deploying teams of heterogeneous robots.
The main questions we want to address in this workshop are: is it possible to mathematically define heterogeneity of the system, what is the role of learning in such systems, what does heterogeneity bring into the table, what are the successful real world examples of these systems and what we can learn from these case studies. This workshop aims to bring together a group of researchers both from academia and industry to discuss these problems. We want to open a conversation between industry practitioners and researchers to highlight the fundamental issues faced in the heterogeneous multi-robot system design and deployment.
University of Michigan
Singapore University of Technology and Design
The University of Virgina
We invite participants to submit their research in IEEE Conference format (up to 4 pages including figures, excluding references). We encourage the submission of early ideas, late-breaking results, position papers, or open research questions that are likely to generate interesting discussions. Accepted papers will be presented in a poster session and selected papers as spotlight talks. All submitted contributions will go through a single blind review process.
The specific goals of the workshop will be to discuss ideas around following topics:
Submissions in PDF accepted at: https://cmt3.research.microsoft.com/icrahmrs2023
IEEE Conference Submission Template: http://ras.papercept.net/conferences/support/support.php
For any questions please constact: firstname.lastname@example.org
University of South Carolina
In-person Venue: South Gallery Room 17, ExCeL London, UK
Remote Participation Link: Zoom link
|9:10||Invited Talk - Prof. Ronald C. Arkin (Georgia Tech)|
|10:00||Invited Talk Dr. Malika Maghjani (Singapore University of Technology and Design)|
|10:35||Coffee break + poster session|
|11:20||Invited Talk - Dr. Andreas Kolling Amazon Robotics|
|11:55||Invited Talk - Danilo Petrocelli (Maritime Robotics)|
|13:30||Invited Talk - Prof. Kazuya Yoshida (Tohoku University)|
|14:05||Invited Talk - Dr. Dimitra Panagou (University of Michigan)|
|14:40||Selected Contributed presentation|
|15:30||Coffee break + poster session|
|16:25||Invited Talk - Dr. Nicola Bezzo (The University of Virginia)|
|17:00||Panel Discussion + closing remarks|
|Invited Talk 1 (May 29, 9:10)|
Heterogeneous Robot Teams: Land, Sea, Air, and Underground Operations
Prof. Ronald C. Arkin
Abstract: At Georgia Tech we have studied robot teaming for over 30 years, starting from multiple robots maintaining behavior-based formations in outdoor settings. We then explored heterogenous air and ground robot teams: from coordinated patrols for situational awareness; to micro autonomous systems in urban environments using conceptual space theory and target discovery and tracking; to broad area maritime surveillance for naval operations; to bio-inspired teams modeled after bird lekking behavior and age-differentiated wolf packs; to multi-robot mission specification and multi-strategy learning; to misdirection/counter-misdirection using robot shills; and most recently underground operations as part of the DARPA Subterranean challenge where we placed second. The evolution of our research moved from teams of similar robots, to heterogeneous teams in differing domains, while recognizing that human-robot teaming is an example of heterogeneous agents while maintaining concern for the societal and ethical implications of this research. This talk surveys a sample of this broad area of our research ranging from behavior based formations, inter-robot communication management, underground task allocation and HRI between the operator and heterogeneous team among others as time permits.
Bio: Professor Ronald C. Arkin is Regents' Professor Emeritus in the College of Computing at Georgia Tech, where he also served as Associate Dean for Research. He served as STINT visiting Professor at KTH in Stockholm, Sabbatical Chair at the Sony IDL in Tokyo, member of the Robotics and AI Group at LAAS/CNRS in Toulouse, and in Brisbane Australia at Queensland University of Technology and CSIRO. Dr. Arkin's research interests include behavior-based control and action-oriented perception for mobile robots and UAVs, deliberative/reactive architectures, robot survivability, multiagent robotics, biorobotics, human-robot interaction, machine deception, robot ethics, and learning in autonomous systems. His books include Behavior-Based Robotics, Robot Colonies, and Governing Lethal Behavior in Autonomous Robots. He has provided expert testimony to the United Nations, the International Committee of the Red Cross, the Pentagon and others on Autonomous Systems Technology. Prof. Arkin served on the Board of Governors of the IEEE Society on Social Implications of Technology, the IEEE Robotics and Automation Society (RAS) AdCom and is a founding co-chair of IEEE RAS Technical Committee on Robot Ethics. He served as a Distinguished Lecturer for the IEEE Society on Social Implications of Technology, is currently a Distinguished Visitor for the IEEE Computer Society and and a Life Fellow of the IEEE.
|Invited Talk 2 (July 1, 10:00)|
Heterogeneous Multi-Robot Systems for Monitoring, Search and Urban Reconnaissance
Dr. Malika Maghjani
Abstract: Real-world applications with multiple objectives can greatly benefit when multi-robot systems are heterogeneous. In this talk, I will present some of the applications for which we use heterogeneous multi-robot systems. These include dynamic environment monitoring, searching for moving targets and urban reconnaissance. The heterogeneity in the proposed applications is showcased in terms of the robots in different domains as well as robots with different capabilities. I will also highlight some of our efforts on learning-based heterogeneous multi-robot systems, the challenges encountered, and the lessons learned.
Bio: Dr. Malika Meghjani is an Assistant Professor in the Computer Science and Design Pillar at Singapore University of Technology and Design (SUTD). She directs the Multi-Agent Robot Vision and Learning (MARVL) Lab, with the focus on algorithm design for efficient, reliable and scalable robots that can work independently and collaboratively with humans. Her research interests are in planning under uncertainty, reinforcement learning, computer vision, deep learning, and game theory. The applications of her work are in field robotics ranging from marine robots specifically, underwater and surface vehicles to aerial drones and self-driving cars as well as other ground vehicles in unstructured environments. Malika has been cited by Analytics Insight in 2020 as one of the World's 50 Most Renowned Women in Robotics. She is also 2017 SMART Postdoctoral Scholar, 2015 McGill Scarlet Key recipient, 2013 IEEE Canada Women in Engineering Prize awardee and 2013 Google Anita Borg Scholar.
|Invited Talk 3 (May 29, 11:20)|
The Value of Heterogenous Autonomous Systems in Industry
Dr. Andreas Kolling
Abstract: Amazon is a pioneer in robotics and has built the largest worldwide fleet of robots, with more than 750,000 robots in continuous operation. Autonomous systems have become a reality in our facilities around the globe from mobile robots to robot arms. Scientists and engineers continue to work on scaling our systems and adding new types of robots. With that new reality come a host of questions around the purpose of autonomy and how to safely and efficiently design robots that interact with millions of people. We will showcase the diversity of these systems, what considerations play a role during design, deployment, and maintenance, and how one might think about the capabilities of a hierarchical and heterogenous system of systems.
Bio: Dr. Andreas Kolling is a principal applied scientist at Amazon Robotics building autonomous systems that are deployed at scale. Previously, he was a principal scientist at iRobot and built the Roomba i7+, which received numerous awards as the most advanced consumer robot. He was an assistant professor at the University of Sheffield and a postdoc at the Robotics Institute at Carnegie Mellon University. His research interests include planning, mapping, multi-robot systems, human-robot interaction and robot software. He has published more than sixty peer-reviewed articles, served as general co-chair for DARS 2016, and as associate editor for ICRA and IROS since 2014.
|Invited Talk 4 (May 29, 11:55)|
Object detection and tracking in Collaborative Multi-Robot settings.
Abstract: Recent technical advancements in both fields of Multi-Robot and Deep Learning Systems have enabled a specific set of applications possible. Multi-Robot Systems are inherently capable of providing greater resilience and versatility in a diverse set of tasks when compared to a single robot platform, owing to their ability to collaborate to achieve multiple objectives. On the other hand, Deep Learning systems can robustly detect and track objects of interest in complex scenarios. However, one of the main problems in the integration of these two areas is the bottleneck of computing Deep Learning applications on resource-limited platforms of UAVs and USVs. To enhance collaborative perception and situational awareness, it is necessary to provide a reliable real-time vision system, capable of operating in a dynamic and complex environment, to deliver useful insights among agents. This talk gives an overview on the ongoing R&D projects at Maritime Robotics. We discuss a general framework that can be adopted in a maritime scenario based on the primary constraints of the resource-limited computing platform of UAVs/USVs.
Bio: Danilo Petrocelli is a Senior Machine Learning Engineer at Maritime Robotics AS. He has been working in the field of computer vision applied to innovative robotics applications for eight years. His career focused on developing machine learning solutions for a broad range of applications such as the development of a Deep Learning-based vision system for Mars rovers to detect anomalies and the design and deployment of DL algorithms to solve complex industrial problems. He is part of the Situational Awareness/Autonomy team at Maritime Robotics AS, where he works on applying Machine Learning solutions contributing to the development of new products in relevant technology.
|Invited Talk 6 (May 29, 13:30)|
Heterogeneous Multi-Robot Systems for Lunar/Planetary Exploration
Prof. Kazuya Yoshida
Abstract: In Space Robotics Lab (SRL) at Tohoku University, Japan, we are developing heterogeneous multi-robot systems for lunar/planetary exploration. So far, robotic exploration missions on the surface of the Moon and remote planets have been conducted by a single capable mobile robot. But if deploying multiple robots, we expect advanced performance in terms of increased coverage areas, accuracy of mapping, adaptability to challenging terrains and robustness to contingent situations, even though the capability of each robot is limited. In this talk, the speaker will introduce the past and current research activities at SRL by highlighting the following topics. (1) Map-free exploration of unknown, unstructured fields with scattered obstacles by multiple robots. A simple algorithm works for the effective expansion of the exploration coverage areas. And the introduction of a macroscopic hierarchy strategy can further increase the performance. (2) Simultaneous mapping and localization (SLAM) are crucial technology for the exploration of the unknown. The idea of loop-closure is well-known to increase the accuracy of the SLAM. And in the scenario of multi-robot exploration, loop-closures among each other robot offer a promising option. Such an approach is termed Co-SLAM. (3) Lastly, an ongoing R&D project for collaborative heterogeneous multi-robot systems for resource exploration and human outpost construction is introduced. Here, the modular robotic design is one of the key points in the development. In space missions where the delivery of new hardware parts and components is not easy, the capability to change the mechanical configuration of the robots by rearranging the modular components offers the self-update of the functionality of the existing robots onsite. This idea will bring robust and sustainable robotics-based activities on the Moon and beyond.
Bio: Professor Kazuya Yoshida received B.Eng., M.Sc. and Dr.Eng, degrees in Mechanical Engineering Science from Tokyo Institute of Technology, Japan, in 1984, 1986, and 1990, respectively. He served as Research Associate of Tokyo Institute of Technology from 1986 to 1994, and Visiting Scientist of Massachusetts Institute of Technology, U.S.A. in 1994. From 1995 to 2003 he was appointed as Associate Professor, and since 2003 he is Full Professor at Space Robotics Lab in Department of Aerospace Engineering, Tohoku University, Japan. His research activities cover dynamics and control of space robotic systems ranging from orbital free-flying robots to planetary exploration rovers. Also, he has been contributing to space robotics education for international students at International Space University in Strasbourg, France (for Master of Space Studies) and various locations around the world (for Summer Study Programs.) Member of IEEE since 1990.
|Invited Talk 5 (May 29, 14:05)|
Trust Adaptation for Multi-Agent Safety via Tunable Control Barrier Functions
Dr. Dimitra Panagou
Abstract: We will present some of our recent results and ongoing work on safety-critical control synthesis under constraints, with applications to heterogeneous multi-agent systems of cooperative and non-cooperative agents. The approach introduces the Rate Tunable Control Barrier Functions, which adapt online the parameters of the safety certificates (Control Barrier Functions) of each agent based on a trust metric/score that each agent builds about its neighbors. The proposed framework aims to eventually develop and integrate adaptive, learning and control methods towards provably-correct and computationally-efficient mission synthesis for multi- agent systems in the presence of constraints and uncertainty.
Bio: Dr. Dimitra Panagou received the Diploma and PhD degrees in Mechanical Engineering from the National Technical University of Athens, Greece, in 2006 and 2012, respectively. In September 2014 she joined the Department of Aerospace Engineering, University of Michigan as an Assistant Professor. Since July 2022 she is an Associate Professor with the newly established Department of Robotics, with a courtesy appointment with the Department of Aerospace Engineering, University of Michigan. Prior to joining the University of Michigan, she was a postdoctoral research associate with the Coordinated Science Laboratory, University of Illinois, Urbana-Champaign (2012-2014), a visiting research scholar with the GRASP Lab, University of Pennsylvania (June 2013, Fall 2010) and a visiting research scholar with the University of Delaware, Mechanical Engineering Department (Spring 2009). Dr. Panagou’s research program spans the areas of nonlinear systems and control; multi-agent systems and networks; motion and path planning; human-robot interaction; navigation, guidance, and control of aerospace vehicles. She is particularly interested in the development of provably- correct methods for the safe and secure (resilient) operation of autonomous systems in complex missions, with applications in robot/sensor networks and multi-vehicle systems (ground, marine, aerial, space). Dr. Panagou is a recipient of the NASA Early Career Faculty Award, the AFOSR Young Investigator Award, the NSF CAREER Award, and a Senior Member of the IEEE and the AIAA.
|14:40-14:48||Steven Ceron, Pascal Spino and Daniela Rus. Heterogeneous Swarms with Reconfigurable Functions and Morphologies.|
|14:48-14:56||Suet Lee and Sabine Hauert. Heterogeneity of Faults in a Robot Swarm: Identifying Discriminatory Metrics.|
|14:56-15:04||Ori A. Miller, Jason M. Gregory, and Christopher Reardon. Teaming Heterogeneous Ground and Micro-Aerial Robots for Following of Non-Cooperative Agents.|
|15:04-15:12||Mohsen Raoufi, Pawel Romanczuk, and Heiko Hamann. Inter-individual Variations in Swarm Robotics with the Case Study of Kilobots.|
|15:12-15:20||Gong Chen, Duong Nguyen-Nam, Malika Meghjani, Phan Minh Tri, Marcel Bartholomeus Prasetyo, Mohammad Alif Daffa, and Tony Q. S. Quek. Heterogeneous Multi-Robot Task Assignment for Urban Reconnaissance.|
|Invited Talk 6 (May 29, 16:25)|
I know that you know that I know: Towards Epistemic Planning of Heterogeneous Robotic Systems
Dr. Nicola Bezzo
Abstract: Thanks to their diversity in capabilities, heterogeneous robotics systems have the potential to be very effective assets for unmanned operations where precise sensing of the surrounding environment and rapid decision making and control are required especially in unknown, unstructured, and possibly hazardous settings. Coordination and allocation of such systems is however a challenging problem due to communication that is often unreliable, causing inefficiencies or outright failures to arise if not properly handled. This talk will discuss such challenges and will present a novel epistemic planning method that allows each robot to operate autonomously in the absence of constant communication: each robot as it is exploring will maintain a belief and empathy of the other teammates and a measure of uncertainty about that belief which will inform subsequent task allocations and rendezvous locations.
Bio: Professor Nicola Bezzo is with the Departments of Systems and Information Engineering (SIE) and Electrical and Computer Engineering (ECE) with courtesy appointment in Computer Science (CS) at the University of Virginia (UVA). Prior to UVA he was a Postdoc at the University of Pennsylvania in the Precise Center working on research related to assured and resilient autonomy. He obtained his PhD in Electrical and Computer Engineering from the University of New Mexico working on multi-robot systems motion planning and control under communication uncertainties and he received both MS and BS summa cum laude from the Politecnico di Milano. At UVA he leads the Autonomous Mobile Robots Lab (AMR Lab) and he is also part and a funding member of the LINK Lab - a recently established CPS center. His research interests are in safe and agile motion planning and control of autonomous robots under uncertainties and resilient and assured autonomy. He has received the Robotics and Automation Magazine best paper award in 2016, the best paper award from the International Conference on Cyber-physical Systems in 2014, and recently the Amazon Faculty Research Award. His research is/has been supported by DARPA, NSF, ONR, and AFRL. His industry collaborators and sponsors include Northrop Grumman, Amazon, Boeing, CoStar, MITRE, Booz Allen Hamilton, and Leidos.
Postdoctoral Associate, UMD
Associate Professor, UMD