Inverse Reinforcement Learning algorithms and features for robot navigation in crowds: An experimental comparison

164 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

End-to-End Robotic Reinforcement Learning without Reward Engineering (2019, background): “However, these methods have limitations of their own: collecting expert demonstrations puts a significant data collection burden on the user, and demonstrations are usually non-intuitive for a person to provide (often involving kinesthetically moving a robot [11] or teleoperating it [39]).”
Learning From Demonstration in the Wild (2018, methodology): “Related to ViBe are several existing LfD methods that learn road and pedestrian behaviour [29], [30], [31], [32].”
Learning socially normative robot navigation behaviors with Bayesian inverse reinforcement learning (2016, methodology): “Those using learned behaviors include [5], [7], [8], in which Henry et al. [5] and Vasquez et al. [7] address the problem of learning to navigate efficiently in crowds of simulated pedestrians.”
Core Challenges of Social Robot Navigation: A Survey (2021, methodology): “[265] recover reward functions describing the motion of humans in a cluttered environment using inverse reinforcement learning (IRL) and use it to inform the costmap of a search-based planner.”
SPENCER: A Socially Aware Service Robot for Passenger Guidance and Help in Busy Airports (2015, unknown): “This is more robust than policy search, because rewards are better generalizable and more succinct (see Vasquez et al, 2014).”