Laser-based gap finding approach to mobile robot navigation

In this paper, a real-time laser based gap finding obstacle avoidance algorithm is presented. This algorithm layers on top of a global planner to maintain the overall goal of a given task. In the presence of an unknown obstacle, the algorithm computes a trajectory toward a gap that is wide enough and is closest to the path pre-planned by the global planner. In order to achieve this result, a four-stage process is executed sequentially namely: data classification, obstacle detection, collision avoidance, and online trajectory generation. During these stages, the algorithm classifies the environment into free space and obstacle regions, adjusts the vehicle velocity as a function of surrounding obstacles proximity, makes a decision to avoid the obstacles and then execute a new trajectory. This trajectory can either be an offset from the original path or a normal path to the best gap depending on the size of the free space, width of the robot and the allowable clearance from obstacles. Experiments show that this approach can avoid obstacles efficiently and effectively achieve the overall goal.