ROBOT NAVIGATION IN VERY COMPLEX, DENSE, AND CLUTTERED INDOOR/OUTDOOR ENVIRONMENTS

This paper addresses the reactive collision-free motion generation for in- door/outdoor robots which have geometric, kinematic, and dynamic constraints. Most of the current mobile robots are designed exhibiting some of these constraints: (1) typically they are circular, square or rectangular robots and (2) they are differential-driven robots, car-like robots, tri-cycle robots, etc. On the other hand, many navigation methods do not take into account the specific shape or robot's kinematics and dynamics. In this case, these methods relax some constraints or they rely on approximations. It is clear that this is a gap in research that needs to be closed, by devising mechanisms to generalize navigation methods to be applied over a wide range of mobile platforms. This paper focuses on the generalization of a reactive method - Nearness Diagram Navigation - to work over a fleet of geometric, kinematic, and dynamic constrained indoor/outdoor mobile robots. This framework has been extensively tested using four indoor and one outdoor robots equipped with different sensors. To validate the method, we report experiments in unknown, non- predictable, unstructured, cluttered, dense and complex environments.