MASc Seminar Notice: "On Legible and Predictable Robot Navigation in Multi-Agent Environments" by Jean-Luc Bastarache

Monday, October 31, 2022 2:00 pm - 2:00 pm EDT

Candidate: Jean-Luc Bastarache
Title: On Legible and Predictable Robot Navigation in Multi-Agent Environments
Date: October 31, 2022
Time: 2:00 pm
Place: via Zoom
Supervisor(s):  Professor Chris Nielsen, Professor Stephen Smith
 

Abstract:

Legibility has recently become an important property to consider in the design of social navigation planners. Legible motion is intent-expressive, which when employed during social robot navigation, allows others to quickly infer the intended avoidance strategy. Predictability, although less commonly studied for social navigation, is, in a sense, the dual notion of legibility, and should also be accounted for in order to promote efficient motions. Predictable motion matches an observer's expectation which, during navigation, allows others to confidently carryout the interaction. In this work, we present a navigation framework capable of reasoning on its legibility and predictability with respect to dynamic interactions, e.g., a passing side. Our approach generalizes the previously formalized notions of legibility and predictability by allowing dynamic goal regions in order to navigate in dynamic environments. This generalization also allows us to quantitatively evaluate the legibility and the predictability of trajectories with respect to navigation interactions. Our approach is shown to promote legible behavior in ambiguous scenarios and predictable behavior in unambiguous scenarios. We also provide an adaptation to the multi-agent case, allowing the robot to reason on its legibility and predictability with respect to multiple interactions simultaneously. This adaptation promotes behaviors that are not illegible to other agents in the environment. In simulation, this is shown to resolve scenarios of high-complexity in an efficient manner. Furthermore, our approach yields an increase in safety while remaining competitive in terms of goal-efficiency when compared to other robot navigation planners in randomly generated multi-agent environments.