Social Navigation

Social navigation aims to help robots move through human-populated spaces while avoiding collisions, respecting personal space, and adapting to people’s behavior. Common approaches combine pedestrian tracking and trajectory prediction with planners such as model predictive control, reinforcement learning, and vision-language policies; datasets and simulations support training and evaluation. Benchmarks increasingly measure social compliance and pedestrian burden alongside robot success, efficiency, and safety. A central challenge is that good behavior depends on context and population, so policies that work in one setting may fail across cultures, crowded spaces, or unexpected interactions.