Robot Motion Planning Based on Improved RRT Algorithm and RBF Neural Network Sliding

Aiming to address the limitations of the traditional rapidly-exploring random tree (RRT) algorithm in robotic arm path planning, such as high randomness, slow planning speed, non-smooth paths, and excessive corners, an improved RRT algorithm incorporating a target bias strategy and artificial potential field approach is proposed. Firstly, this algorithm employs a probabilistic sampling strategy to facilitate quicker growth of the tree structure towards the target point, thus accelerating the planning speed of the RRT algorithm. Subsequently, a repetitive greedy strategy is introduced to reduce redundant nodes in the path and enhance search efficiency. Furthermore, an artificial potential field combined with a target bias strategy is integrated to guide the RRT algorithm towards the target direction. Finally, a third-order B-spline curve optimization is introduced to ensure smoother paths. Compared to traditional RRT algorithms and RRT* algorithms, the improved RRT algorithm exhibited an increase in planning speed of 46.1% and 27.0%, 41.8% and 20.9% for two-dimensional and three-dimensional environments respectively. Path lengths were reduced by 20.9% and 10.6%, 22.0% and 8.7%, while the number of search nodes notably decreased. Considering the complexity of robotic arm operations and uncertainties in system parameters that hinder the establishment of accurate mathematical modeling, a sliding mode control based on radial basis function (RBF) neural networks is proposed. To substantiate the enduring stability of the designed controller, we harness the Lyapunov stability principle. The algorithm validation for path planning and trajectory tracking of the robotic arm are conducted in the MATLAB simulation environment. Experimental results illustrate significant improvements in terms of iteration count, planning speed, path length, and smoothness of the RRT algorithm. Moreover, the motion trajectory of the robotic arm is basically consistent with the expected trajectory.