KINODYNAMIC RRT PLANNING WITH BACKSTEPPING TRACKING FOR CONSTRAINED MOBILE ROBOT MOTION IN CLUTTERED ENVIRONMENTS

Tóm tắt

Autonomous mobile robots often require motion plans that are not only collision-free but also consistent with velocity and acceleration limits that matter during execution. This paper studies a coupled planning-and-tracking pipeline for a planar second-order robot model. A kinodynamic RRT* planner generates reference trajectories by forward propagation under bounded acceleration and speed, and a backstepping controller tracks the resulting reference using the planner-provided acceleration signal. The approach is evaluated in simulation on three obstacle-field maps, including a dense layout and a narrow-passage scenario, using 30 randomized runs per map. Compared with a kinodynamic RRT baseline without rewiring, kinodynamic RRT* reduces trajectory cost and path length while increasing computation time due to neighborhood search and rewiring. On the same planned references, the backstepping tracker produces smaller average tracking errors than a tuned PID controller, with mean position and velocity errors of 0.042m and 0.104m/s, respectively. The results illustrate how aligning the planning model and the tracking design can improve repeatability of closed-loop motion in cluttered environments, while keeping the overall computation practical for offline planning and low-rate replanning.