面向异构环境自主巡航的仿人机器人运动规划及多足平台推广研究

In order to realize the autonomous navigation in heterogeneous environment for multi-legged platform represented by humanoid robot, this project tries to resolve the difficulties of 3 independent models tackling motion planning, namely continuous path model, discrete footstep model and whole-body model. It combines these models to establish a hybrid one. It as well, based on region partitioning of heterogeneous environment, introduces the integrated planning strategy, improves the local path following performance, and finally realizes the planning with high efficiency. It tries to establish the principal of region partitioning with 3D environment model decomposition and synthesis, and realizes the integration and optimization of multi strategies with iterative anytime incremental thought; it also introduced the omni-directional moving vehicle model to take account of inter-step connectivity, and treats multi-legged robots as humanoid for generalization of the planning method. With dynamics simulation and real robot experiment in the framework of navigation systems, this project tries to prove the validity of the method and generalizes it to the field of legged robots, and reaches the goal of improving the environment adaptivity with real-time decision-making and high speed moving ability.

以实现仿人机器人为代表的关节式多足平台在异构环境下的自主巡航为目标，针对连续路径、离散足迹和全身关节等规划模型难以单独解决复杂运动规划的问题，将其综合形成混合移动模型，在异构环境区域分割的基础上，构建综合规划策略，提升局部路径跟随速度，实现高效规划。拟通过三维环境的多视图分解和综合，构建区域分割的系统性方法，以实时修正增量式迭代策略实现多种规划方法的融合与综合优化，引入全方位小车模型考虑步幅间连接约束实现快速路径跟随，发掘多足机器人与双足机器人的共性实现方法的推广。通过动力学仿真和机器人样机在自主巡航整体框架下的综合实验证实方法的有效性，进而推广至普遍的关节式足式机器人。从而达到综合提升关节足式机器人环境适应性、决策实时性和移动快速性的目标。

以实现双足步行仿人机器人为代表的足式机器人在异构环境下的高效自主巡航为目标，针对连续路径、离散足迹和全身关节等规划模型难以单独解决复杂运动规划的问题，综合形成混合移动模型，在异构环境区域分割的基础上，构建综合规划策略，提升局部路径跟随速度，实现高效规划。依赖激光雷达结合二自由度云台获得复杂环境的高度信息，采用2D地图配合3D点云、欧式距离变换、平面地图轮廓查找等一系列处理流程获得了开放空间与离散落地空间的集合，构建了区域分割的系统性方法。通过多个运动基元前后连接构成路径的平均足迹数量，以及运动基元间的过渡时间以及运动基元的总数，获得了连续模型下的等价代价。利用连续路径规划下的运动基元与离散足迹规划下的可行落地足区域的相似性，针对连续路径区、离散足迹区以及两者之间的过渡区，选择性的拓展连续路径运动基元和离散落地足迹运动基元，实现了复杂异构环境下规划算法的协同优化。以实时修正增量式迭代策略实现多种规划方法的融合与综合优化，引入全方位小车模型考虑步幅间连接约束实现快速路径跟随。同时构建了动力学仿真平台和全新的MOS-7机器人样机，在自主巡航整体框架下的综合实验证实方法的有效性，从而达到综合提升关节足式机器人环境适应性、决策实时性和移动快速性的目标。