Research on an Improved Informed RRT*-Connect Algorithm for Path Planning

Path planning is a core challenge in robotics, especially for robotic arms operating in high-dimensional configuration spaces. In complex environments, efficiently finding an optimal path for a robotic arm remains a critical problem. The Informed RRT*-Connect algorithm enhances path planning efficiency by employing an informed sampling strategy. After obtaining the first feasible solution, this strategy restricts the sampling range to valid regions of the solution space, thereby reducing redundant sampling points. To further adapt the algorithm to high-dimensional and complex scenarios while improving convergence speed, this paper proposes an enhanced path planning algorithm-Informed-Obstacle RRT*-Connect. Building on the informed sampling strategy, the proposed algorithm integrates an obstacle-based sampling strategy, enabling faster convergence to potential optimal solution regions at different stages. Additionally, a variable-step strategy and a goal-biased strategy are introduced to accelerate the growth and connection of the start and goal trees, significantly improving computational efficiency. Experimental results across various map environments demonstrate that the proposed algorithm outperforms the original Informed RRT*-Connect in both success rate and computation time for optimal solutions, while also exhibiting higher stability and robustness. Ultimately, this research provides the theoretical foundation for applying this algorithm to robotic arm path planning, with future implementation and validation to be conducted within the ROS2 platform.