基于最小snap的机器人自主探索路径规划算法

Robot autonomous discovery path planning algorithm based on minimum snap

基于最大化信息增益的自主探索路径规划算法在小规模场景下探索效率高,但是由于其缺乏全局性在复杂场景中探索效率差,并且探索轨迹不平滑,机器人无法直接追踪.通过分层思想将路径规划分为全局与局部两部分,使用全局规划结果引导局部探索规划,保证了局部探索与全局探索的一致性.基于观察点选择过程中的子模性,提出引入随机性的观察点选择算法,保证了观察点选择的鲁棒性.将路径规划问题分解为观察点访问顺序选择和轨迹平滑问题,通过求解旅行商问题确定访问顺序,通过最小snap方法平滑探索轨迹,实现了高效的自主探索路径规划算法.仿真实验结果表明,基于最小snap的探索路径规划算法在复杂场景中的探索效率相比其他算法更高,并拥有良好的算法复杂度,可以保证机器人自主探索复杂场景.

The autonomous exploration path planning algorithm based on maximizing information gain has high exploration efficiency in small-scale scenes,but due to it was lacked of globality,the exploration efficiency was poor in complex scenes,and the exploration trajectory was not smooth,and the robot cannot track directly.The path planning was divided into global and local parts through the hierarchical idea,and the global planning results were used to guide the local exploration planning,so as to ensure the consistency of local exploration and global exploration.Based on the submodularity in the process of observation point selection,a random observation point selection algorithm was proposed to ensure the robustness of observation point selection.The path planning problem was decomposed into the observation point access order selection and trajectory smoothing problem.The access order was determined by solving the traveling salesman problem,and the exploration trajectory was smoothed by the minimum snap method,which realized an efficient autonomous exploration path planning algorithm.The simulation results showed that the exploration path planning algorithm based on minimum snap has higher exploration efficiency in complex scenes than other algorithms,and has good algorithm complexity,which can ensure the robot to explore complex scenes independently.