Zhang Bo, Lu Shanlong, Li Qing, Du Peng, Hu Kaixin
School of Automation, Beijing Information Science and Technology University, Beijing 100192, China.
International Research Center of Big Data for Sustainable Development Goals, Beijing 100094, China.
Sensors (Basel). 2024 Nov 28;24(23):7596. doi: 10.3390/s24237596.
To address the design and application requirements for USVs (Unmanned Surface Vehicles) to autonomously escape from constrained environments using a minimal number of sensors, we propose a path planning algorithm based on the RRT* (Rapidly Exploring Random Tree*) method, referred to as BN-RRT* (Blind Navigation Rapidly Exploring Random Tree*). This algorithm utilizes the positioning information provided by the GPS onboard the USV and combines collision detection data from collision sensors to navigate out of the trapped space. To mitigate the inherent randomness of the RRT* algorithm, we integrate the Artificial Potential Field (APF) method to enhance directional guidance during the sampling process. Additionally, inspired by blind navigation principles, we propose an active collision mechanism that relies on continuous collisions to identify obstacles and adjust the next movement direction, thereby improving the efficiency of escape path planning. We also implement an obstacle memory mechanism to prevent exploration into erroneous areas during sampling, significantly increasing the success rate of escape and reducing the path length. We validate the proposed algorithm in a dedicated MATLAB environment, comparing its performance with existing RRT, RRT*, and APF-RRT* algorithms. Experimental results indicate that the improved algorithm achieves significant enhancements in both planning speed and path length compared to the other methods.
为满足无人水面舰艇（USV）使用最少数量传感器自主逃离受限环境的设计和应用要求，我们提出了一种基于RRT*（快速扩展随机树*）方法的路径规划算法，称为BN - RRT*（盲导航快速扩展随机树*）。该算法利用USV上GPS提供的定位信息，并结合来自碰撞传感器的碰撞检测数据，以驶出被困空间。为减轻RRT算法固有的随机性，我们整合了人工势场（APF）方法，以在采样过程中增强方向引导。此外，受盲导航原理启发，我们提出了一种主动碰撞机制，该机制依靠连续碰撞来识别障碍物并调整下一个移动方向，从而提高逃生路径规划的效率。我们还实现了一种障碍物记忆机制，以防止在采样过程中探索错误区域，显著提高逃生成功率并缩短路径长度。我们在专用的MATLAB环境中验证了所提出的算法，并将其性能与现有的RRT、RRT和APF-RRT*算法进行了比较。实验结果表明，与其他方法相比，改进后的算法在规划速度和路径长度方面都有显著提升。
