Self-Navigation in Non-Adversarial Crowds: An Invariant Set-Based Approach

Self-navigation in noncoordinating crowded environments is a challenging task within multiagent systems consisting of nonholonomic robots operating through local sensing.Here,self-navigationrefers to complete autonomy without relying on any external systems for navigation. In this setting, our primary objective is the development of a fast, novel, sensor-driven controller that directly computes control commands to enable safe maneuvering while coexisting with other agents. We propose an input-constrained feedback controller for nonholonomic mobile robots and characterize the associated invariant sets. These invariant sets are the key to maintaining stability and safety amidst noncooperating agents. We then introduce a planning strategy that steers the generation of invariant sets toward the agent’s intended target, enabling robots to compute provably safe control inputs without preplanned paths or trajectories when navigating crowded multiagent environments. The practicality of our technique is demonstrated through hardware experiments, and we provide a comparison with existing navigation strategies where each robot makes independent sensing and decisions. The proposed approach is well-suited to scenarios that demand rapid, on-the-fly control computations based solely on real-time perceived safety bounds.