To guide an unmanned aerial vehicle(UAV)flying in complex three-dimensional(3D)environments with unknown obstacles,a novel UAV path planning algorithm named IRRT^(∗)-C2TD3 is proposed.The algorithm combines the rapidl...To guide an unmanned aerial vehicle(UAV)flying in complex three-dimensional(3D)environments with unknown obstacles,a novel UAV path planning algorithm named IRRT^(∗)-C2TD3 is proposed.The algorithm combines the rapidly-exploring random tree star(RRT^(∗))algorithm with the twin delayed deep deterministic policy gradients(TD3)algorithm(a deep reinforcement learning algorithm).By employing exploration strategies from reinforcement learning,IRRT^(∗)-C2TD3 improves the RRT^(∗)algorithm.IRRT^(∗)-C2TD3 is a two-stage path planning algorithm comprising pre-planning and real-time planning.It performs pre-planning of paths by generating paths based on geometric connections toward the goal and smoothing them using cubic B-spline curves.By designing the network architecture and reward function of the TD3 algorithm,real-time planning in unknown environments is achieved based on the pre-planned path from the first stage.Simulation results show that IRRT^(∗)-C2TD3 demonstrates better path planning performance in 3D partially unknown environments than RRT^(∗)-C2TD3,M-C2TD3 and MODRRT^(∗)algorithms.展开更多
针对复杂水下环境中水下自主航行器(autonomous underwater vehicle,AUV)路径规划问题,提出一种改进启发式快速随机扩展树(rapidly-exploring random trees,RRT)的路径规划算法。针对路径点采样过程中缺乏目标导向性的问题,采用目标点...针对复杂水下环境中水下自主航行器(autonomous underwater vehicle,AUV)路径规划问题,提出一种改进启发式快速随机扩展树(rapidly-exploring random trees,RRT)的路径规划算法。针对路径点采样过程中缺乏目标导向性的问题,采用目标点概率偏置采样策略与目标偏向扩展策略,可使目标节点在随机采样时成为采样点。在路径点扩展过程中,使非目标采样点的扩展结点位置偏向于目标点的方向,从而增强算法在随机采样与扩展过程中的目标搜索能力。为解决水下路径规划过程中存在过多无效搜索空间的问题,在随机采样过程中引入启发式采样策略,构建包含所有初始路径的采样空间子集,减小采样空间范围,从而提高算法的空间搜索效率。针对AUV在水下环境中抗洋流扰动能力不足的问题,采用速度矢量合成法,使AUV速度矢量与洋流速度矢量合成后指向期望路径的方向,从而抵消水流的影响。在山峰地形中叠加多个Lamb涡流模拟水下流场环境,进行多次仿真实验。实验结果表明:改进启发式RRT算法解决了采样过程中随机性问题,显著缩小了搜索空间,兼顾了路径的安全性与平滑性,并使AUV具有良好的抗洋流扰动能力。展开更多
Dynamic path planning is crucial for mobile robots to navigate successfully in unstructured envi-ronments.To achieve globally optimal path and real-time dynamic obstacle avoidance during the movement,a dynamic path pl...Dynamic path planning is crucial for mobile robots to navigate successfully in unstructured envi-ronments.To achieve globally optimal path and real-time dynamic obstacle avoidance during the movement,a dynamic path planning algorithm incorporating improved IB-RRT∗and deep reinforce-ment learning(DRL)is proposed.Firstly,an improved IB-RRT∗algorithm is proposed for global path planning by combining double elliptic subset sampling and probabilistic central circle target bi-as.Then,to tackle the slow response to dynamic obstacles and inadequate obstacle avoidance of tra-ditional local path planning algorithms,deep reinforcement learning is utilized to predict the move-ment trend of dynamic obstacles,leading to a dynamic fusion path planning.Finally,the simulation and experiment results demonstrate that the proposed improved IB-RRT∗algorithm has higher con-vergence speed and search efficiency compared with traditional Bi-RRT∗,Informed-RRT∗,and IB-RRT∗algorithms.Furthermore,the proposed fusion algorithm can effectively perform real-time obsta-cle avoidance and navigation tasks for mobile robots in unstructured environments.展开更多
针对传统RRT(rapidly exploring random tree)算法在复杂环境下收敛速度慢、存在重复采样、缺乏目标导向性和规划的路径质量不高的问题,提出一种贪婪搜索和目标导向的RRT算法(RRT-D算法),在传统RRT算法的基础上,改进节点的采样方式和父...针对传统RRT(rapidly exploring random tree)算法在复杂环境下收敛速度慢、存在重复采样、缺乏目标导向性和规划的路径质量不高的问题,提出一种贪婪搜索和目标导向的RRT算法(RRT-D算法),在传统RRT算法的基础上,改进节点的采样方式和父节点的选取策略,取消步长限制,通过贪婪式的搜索方式一次生长10个候选节点,选取符合条件的且距离目标点最近的候选点作为子节点生长到树中,提高了算法的搜索能力,降低了路径代价;用动态减少重复搜索区域的方式减少了无效搜索;每次采样后判断采样点能否与目标点直接相连,增加了采样的目标导向性,提高了搜索效率,遍历全树构成无向图时,可根据总采样点数量,通过限制无向图边的长度来减少边的数量,由Dijkstra算法搜索代价最小的路径;最后由分段三次Hermite插值函数对路径进行平滑处理。试验结果表明,与传统RRT算法相比,RRT-D算法不仅大幅缩短了规划时间,而且得到的路径代价更小、更加平滑,节点的利用率更高,验证了RRT-D算法在路径规划中的优势。展开更多
针对快速搜索随机树(rapidly-exploring random tree,RRT)算法的随机采样特征导致的收敛速度慢、路径冗余度高、采样点利用率低问题,给出一种新的解决方法。首先,根据图复杂度公式,计算出图的复杂度后确定目标偏执概率,建立偏置概率自...针对快速搜索随机树(rapidly-exploring random tree,RRT)算法的随机采样特征导致的收敛速度慢、路径冗余度高、采样点利用率低问题,给出一种新的解决方法。首先,根据图复杂度公式,计算出图的复杂度后确定目标偏执概率,建立偏置概率自适应模型;其次,在首次规划好路线后,路径中仍存在一些不必要的拐点与棱角,针对传统路径裁剪依赖局部搜索策略,可能导致次优解生成,提出PRM-Dijkstra(probabilistic roadmap-dijkstra)算法对路径进行裁剪,将改进RRT算法生成的树节点利用PRM算法相互连接起来,通过Dijkstra算法计算出一条最优路径;最后,改进RRT算法与PRM-Dijkstra种算法优势相结合,在保证有一条路径的前提下,最大概率的寻找最优路径。通过复杂图下仿真避障实验,结果显示:改进RRT算法在节点生成数量与规划用时相较传统RRT算法平均减少80%,相较于Goal-bias RRT算法均减少40%。并通过机器人操作系统(robot operating system,ROS)下的MoveIt!集成开发平台进行现实环境下避障实验,验证了算法的可行性与有效性。展开更多
基金National Natural Science Foundation of China(No.62173084)Foundation of Shanghai Committee of Science and Technology,China(Nos.23ZR1401800 and 22JC1401403)。
文摘To guide an unmanned aerial vehicle(UAV)flying in complex three-dimensional(3D)environments with unknown obstacles,a novel UAV path planning algorithm named IRRT^(∗)-C2TD3 is proposed.The algorithm combines the rapidly-exploring random tree star(RRT^(∗))algorithm with the twin delayed deep deterministic policy gradients(TD3)algorithm(a deep reinforcement learning algorithm).By employing exploration strategies from reinforcement learning,IRRT^(∗)-C2TD3 improves the RRT^(∗)algorithm.IRRT^(∗)-C2TD3 is a two-stage path planning algorithm comprising pre-planning and real-time planning.It performs pre-planning of paths by generating paths based on geometric connections toward the goal and smoothing them using cubic B-spline curves.By designing the network architecture and reward function of the TD3 algorithm,real-time planning in unknown environments is achieved based on the pre-planned path from the first stage.Simulation results show that IRRT^(∗)-C2TD3 demonstrates better path planning performance in 3D partially unknown environments than RRT^(∗)-C2TD3,M-C2TD3 and MODRRT^(∗)algorithms.
文摘针对复杂水下环境中水下自主航行器(autonomous underwater vehicle,AUV)路径规划问题,提出一种改进启发式快速随机扩展树(rapidly-exploring random trees,RRT)的路径规划算法。针对路径点采样过程中缺乏目标导向性的问题,采用目标点概率偏置采样策略与目标偏向扩展策略,可使目标节点在随机采样时成为采样点。在路径点扩展过程中,使非目标采样点的扩展结点位置偏向于目标点的方向,从而增强算法在随机采样与扩展过程中的目标搜索能力。为解决水下路径规划过程中存在过多无效搜索空间的问题,在随机采样过程中引入启发式采样策略,构建包含所有初始路径的采样空间子集,减小采样空间范围,从而提高算法的空间搜索效率。针对AUV在水下环境中抗洋流扰动能力不足的问题,采用速度矢量合成法,使AUV速度矢量与洋流速度矢量合成后指向期望路径的方向,从而抵消水流的影响。在山峰地形中叠加多个Lamb涡流模拟水下流场环境,进行多次仿真实验。实验结果表明:改进启发式RRT算法解决了采样过程中随机性问题,显著缩小了搜索空间,兼顾了路径的安全性与平滑性,并使AUV具有良好的抗洋流扰动能力。
基金the National Natural Science Foundation of China(No.61973275)。
文摘Dynamic path planning is crucial for mobile robots to navigate successfully in unstructured envi-ronments.To achieve globally optimal path and real-time dynamic obstacle avoidance during the movement,a dynamic path planning algorithm incorporating improved IB-RRT∗and deep reinforce-ment learning(DRL)is proposed.Firstly,an improved IB-RRT∗algorithm is proposed for global path planning by combining double elliptic subset sampling and probabilistic central circle target bi-as.Then,to tackle the slow response to dynamic obstacles and inadequate obstacle avoidance of tra-ditional local path planning algorithms,deep reinforcement learning is utilized to predict the move-ment trend of dynamic obstacles,leading to a dynamic fusion path planning.Finally,the simulation and experiment results demonstrate that the proposed improved IB-RRT∗algorithm has higher con-vergence speed and search efficiency compared with traditional Bi-RRT∗,Informed-RRT∗,and IB-RRT∗algorithms.Furthermore,the proposed fusion algorithm can effectively perform real-time obsta-cle avoidance and navigation tasks for mobile robots in unstructured environments.
文摘针对传统RRT(rapidly exploring random tree)算法在复杂环境下收敛速度慢、存在重复采样、缺乏目标导向性和规划的路径质量不高的问题,提出一种贪婪搜索和目标导向的RRT算法(RRT-D算法),在传统RRT算法的基础上,改进节点的采样方式和父节点的选取策略,取消步长限制,通过贪婪式的搜索方式一次生长10个候选节点,选取符合条件的且距离目标点最近的候选点作为子节点生长到树中,提高了算法的搜索能力,降低了路径代价;用动态减少重复搜索区域的方式减少了无效搜索;每次采样后判断采样点能否与目标点直接相连,增加了采样的目标导向性,提高了搜索效率,遍历全树构成无向图时,可根据总采样点数量,通过限制无向图边的长度来减少边的数量,由Dijkstra算法搜索代价最小的路径;最后由分段三次Hermite插值函数对路径进行平滑处理。试验结果表明,与传统RRT算法相比,RRT-D算法不仅大幅缩短了规划时间,而且得到的路径代价更小、更加平滑,节点的利用率更高,验证了RRT-D算法在路径规划中的优势。
