[hal-05390292] A Versatile Heuristic-based Social Robot Navigation Solution Efficient in Sparse and Crowded Environments

Human-aware robotic navigation remains an open challenge, with no universal solution yet available. Recent advances in the field have identified key metrics and scenarios that such navigation systems should address. In this paper, we present a novel social robotic navigation solution based on a heuristic approach that yields good performance in various environments (open areas or corridors) and different conditions (absence of human to crowded). Simulations first show the lack of generalization of existing methods and that our approach compete with baselines in crowded scenarios and outperforms them across daily scenarios (corridors), using a well-established set of metrics in the field. Preliminary real-world experiment further demonstrates the feasibility of deploying this approach in dynamic environments involving humans.