Abstract: Aiming at the problems of low efficiency of traditional RRT-Connect algorithm in path planning and poor dynamic obstacle avoidance in dense and complex environment, an improved RRT-Connect and DWA fusion algorithm is proposed. The algorithm guides the growth of random trees by improving sampling strategy, dynamic step optimization and collision detection. Greedy strategy and angle constraint are used to optimize paths in random trees. Based on the kinematics model of the inspection robot, the trajectory cluster is generated through the velocity sampling space. The fuzzy logic system adaptively adjusts the weight coefficient of the evaluation function of DWA algorithm, and integrates the global optimal path point into DWA algorithm to realize the global optimal path and real-time obstacle avoidance. The simulation results show that in the dense and complex environment of oil and gas station, the improved RRT-Connect algorithm shorths the path by 27.09%, improves the smoothness by 84.6%, and increases the collision distance by 18.75% compared with the traditional algorithm. The improved fusion algorithm reduces the path by about 2.97%, increases the smoothness by about 78.8%, and increases the collision distance by about 30.6%, which verifies the effectiveness of the proposed algorithm.

Key words: path planning, improved RRT-Connect algorithm, dynamic window approach (DWA) algorithm, fusion algorithm

摘要： 针对传统RRT-Connect算法在密集复杂环境中路径规划效率低、动态避障效果差等问题，提出一种改进RRT-Connect与DWA融合算法。该算法通过改进采样策略、动态步长优化和碰撞检测引导随机树生长；在随机树中采用贪心策略和角度约束优化路径。基于巡检机器人建立运动学模型，通过速度采样空间生成轨迹簇；建立模糊逻辑系统自适应调整DWA算法评价函数的权重系数，将全局最优路径点融入DWA算法中实现全局最优路径和实时避障。仿真结果表明，在油气站场密集复杂环境中，改进RRT-Connect算法较传统算法路径缩短约27.09%，平滑度提高约84.6%，碰撞距离提高约18.75%；改进融合算法路径减少约2.97%，平滑度提高约78.8%，碰撞距离提高约30.6%，验证了提出算法的有效性。

关键词: 路径规划, 改进RRT-Connect算法, DWA算法, 融合算法