We recently published a new article about an improved version of our Social Environment for Autonomous Navigation. The article is entitled “SEAN 2.0: Formalizing and Generating Social Situations for Robot Navigation” and proposes to describe navigation contexts based on social situations, which encompass the robot task and environmental factors (including nearby pedestrian behavior). In addition, the paper proposes five logic classifiers for common social situations that help understand navigation performance across social contexts, and an approach to generate crowd motion that facilitates simulating these situations in practice with SEAN 2.0. Congratulations to all the authors! Thanks also to Amazon for supporting this research.
Recently, the SEAN team also organized a workshop at ICRA’22 on Social Robot Navigation: Advances and Evaluation in collaboration with researchers at CMU. As part of this event, the workshop organizers launched a new social robot navigation benchmark called SEANavBench. The benchmark remains open for submissions through benchmark.interactive-machines.com.