Making Real-World Multi-Robot Systems More Practical

Add to your list(s) Download to your calendar using vCal

• Dr Ajay Shankar, Computer Laboratory, Cambridge
• Tuesday 12 December 2023, 12:00-13:00
• Institute of Manufacturing (IfM), Seminar Room 1.

If you have a question about this talk, please contact Slawomir Tadeja.

Abstract:

There is a leap to be made when going from multi-‘agent’ to multi-‘robot’ problems: the reality of our systems and components (such as locomotion, sensing, inter-agent communication, etc) kicks in and undercuts several theoretical developments we otherwise rely upon. Complete system integration is thus key to realising multi-robot teams that are robust and resilient in practice, and this requires careful design choices not only in the engineering aspects but also in the theoretical abstractions we choose for them. For instance, lots of work in multi-agent coordination assumes synchronous communications (which is completely impractical beyond lab spaces), and perhaps a better architecture might prevent agents from being too dependent on it (and retain a certain degree of autonomy when comms fail). In this talk, I will present two recent areas of work by which we have tackled and relaxed some of the assumptions implicitly made in multi-robot problems. The first of these embeds diversity in multi-robot teams (with reinforcement learning) and thus makes them resilient in practice. The second allows two airborne multirotor systems to dock with each other by efficiently learning to compensate for aerodynamic effects.

Speakers Bio:

I am a post-doc in the Department of Computer Sc. and Technology at the University of Cambridge (UK), and I work with Dr Amanda Prorok at the ProrokLab (proroklab.org). I completed my PhD at the University of Nebraska-Lincoln (USA), where I was at the Nimbus Lab (nimbus.unl.edu) and developed aerial robotic systems that make tactile interactions with dynamic objects in their surroundings as well as with each other in the air. My work is driven mostly by field robotics and applications, and I also develop full-stack systems that are meant for studying and monitoring our natural environment. In my recent work, I have been more interested in developing real-world systems at larger scales that still operate with a high degree of autonomy and robustness. Please find more at https://www.cl.cam.ac.uk/~as3233/

This talk is part of the Cyber-Human Lab Seminar Series series.

This talk is included in these lists:

• Cyber-Human Lab Seminar Series
• Institute of Manufacturing (IfM), Seminar Room 1

Note that ex-directory lists are not shown.