Real-Time Hybrid Navigation System-Based Path Planning and Obstacle Avoidance for Mobile Robots

In this work, we present a complete hybrid navigation system for a two-wheel differential drive mobile robot that includes static-environment- global-path planning and dynamic environment obstacle-avoidance tasks. By the given map, we propose a multi-agent A-heuristic algorithm for finding the optimal obstacle-free path. The result is less time-consuming and involves fewer changes in path length when dealing with multiple agents than the ordinary A-heuristic algorithm. The obtained path was smoothed based on curvature-continuous piecewise cubic Bézier curve (C2 PCBC) before being used as a trajectory by the robot. In the second task of the robot, we supposed any unforeseen obstacles were recognized and their moving frames were estimated by the sensors when the robot tracked on the trajectory. In order to adapt to the dynamic environment with the presence of constant velocity obstacles, a weighted-sum model (WSM) was employed. The 2D LiDAR data, the robot’s frame and the detected moving obstacle’s frame were collected and fed to the WSM during the movement of the robot. Through this information, the WSM chose a temporary target and a C2 PCBC-based subtrajectory was generated that led the robot to avoid the presented obstacle. Experimentally, the proposed model responded well in existing feasible solution cases with fine-tuned model parameters. We further provide the re-path algorithm that helped the robot track on the initial trajectory. The experimental results show the real-time performance of the system applied in our robot.