Obstacle Avoidance for Autonomous Mobile Robots in Unstructured Human Environments

This paper presents an obstacle avoidance approach for autonomous mobile robots in unstructured human environments, focusing on avoiding irregular obstacles that cannot be detected by traditional 2D laser-based obstacle detection techniques. With the aid of ultrasound sensors and RGB-D cameras, the developed approach can effectively avoid obstacles below the laser scanning plane, which can address the challenges of mobile robots crashing human's feet in unstructured human environments. Furthermore, the developed approach can reliably avoid obstacles hanging in the air, which put up severe challenges to mobile robots working in such unstructured human environments as airports. The effectiveness of the developed obstacle avoidance approach is verified by extensive experiments.