Mobile robot navigation amidst humans with intents and uncertainties: A time scaled collision cone approach

Abstract— We propose a novel collision avoidance formulation in the intent space, suitable for navigation of non-holonomic robots in human centered environments. The intent space is characterized by various bands of trajectories wherein each band can be thought to be a representation of a possible human intended motion and the uncertainty associated with it. We ascribe probabilities to human intentions and characterize the uncertainty around it through Gaussian state transition and its concomitant Gaussian parametric distribution. Given an intent space we design avoidance maneuvers based on our recent works on time scaled collision cone concept which provides analytical characterization of collision free velocities in dynamic environments. In this paper, we present a probabilistic variant of the time scaled collision cone which allows us to relate space of collision free velocities to an associated confidence measure. We also develop an optimization framework which extract such spec...

• Type: Journal
• Year: 2015
• Where: CDC
• Authors: Akhil Nagariya, Bharath Gopalakrishnan, Arun Kumar Singh, Krishna Gupta, K. Madhava Krishna
• Keywords: CDC 2015, Control Systems