earticle

영어

Navigation can be defined as moving a robot to a defined objective and avoiding undesirable situations (collision and unsafe conditions) simultaneously. In this work, a navigation technique based on Potential Field has been used to plan an optimal-safe path to pilot a mobile robot in known environment. The proposed techniques are based on extracting path by merging information from two potential maps: security potential, pulling back robot from obstacles, and slope potential attracting robot to target. Three planning techniques are proposed; weighting technique that gives equal importance to safety and nearest path yet changes importance to highlight attraction in conflict situations. Degraded technique, which is more favorable for cases where security is not often important, so superior weight is given to nearest path. The last technique is to integrate paths from the two previous techniques to have safe paths in some situations and have nearest path in other situations. Comparison between proposed techniques with addition to previous work (graph-based technique) has been done to discuss difference and possible optimizations. The proposed techniques have shown interesting results especially for complex environments with curved obstacles, which is not resolvable with graph based technique. Tests and implementation has been done using Matlab simulations and a real mobile robot with a bird's-eye view camera to discuss techniques and propose future work.