面向复杂地形的四足机器人脚掌建模与控制策略研究

As parts that directly contact with the external environment during the movement of quadruped robots, proper structure and control scheme of the footpads are vital to motion stability. For robots designed for different situations, different footpad structures are required. However, the investigation on quadruped robots so far mainly focuses on the trajectory planning of legs and configuration control of the torso while the specific study of footpads of quadruped robots is not so frequently found. In this project, the structures and control approaches of the footpads of quadruped robots will be investigated for the movement on unstructured environment, which mainly includes the following contents: 1) the structure and mathematic model of the footpads, aiming at mechanical structure analysis and mathematic model establishment; 2) research on the control approaches of the footpads based on compound impedance control, aiming at the optimal compound plan to realize the control of the joints of footpads; 3) research on foot-ground interaction with torso attitude constraint, aiming at the mathematic model establishment for the robot as a whole interacts with the environment to realize the quantitively analysis to the stability in foot-ground interaction. 4) Classified compliant control in unstructured environment based on motion model decision using multiple motion parameters, aiming at improving accuracy of the motion model decision and enhance the compliance of the torso. This project will be beneficial to improving the adaptivity and compliance of the torso.

脚掌作为四足机器人运动过程中与地面直接接触的部位，对机器人运动的稳定性起到至关重要的作用。工作于不同环境下的机器人的脚掌结构也不相同。然而，目前四足机器人的相关研究主要集中在腿部轨迹规划与躯干位姿控制上，而对于四足机器人脚掌的针对性研究相对不足。本项目针对四足机器人在复杂地形中的运动模式，拟对四足机器人的脚掌结构与控制方式进行研究，主要包括：1) 脚掌拓扑结构与数学模型，实现脚掌的结构分析与力学模型建立；2) 基于复合阻抗控制的脚掌控制方式研究，以寻找最优复合方式，实现脚掌关节的控制；3) 基于躯干姿态约束的足地交互研究，建立四足机器人与地面交互的整体数学模型，实现足-地交互过程的定量分析；4) 基于多参数运动模式决策与足-腿协同的分层柔顺控制，借助脚掌与腿部的协同运动，提高机器人运动模式决策的准确性与躯干运动的柔顺性。该项目的研究成果将有助于提高四足机器人在非结构环境中的适应性与柔顺性。

四足机器人的脚掌以及腿部柔顺控制是机器人领域的研究热点问题之一。对于常规运动而言，相关研究相对成熟，很多技术难点均有所突破，而在非常规运动中，四足机器人的运动稳定性以及大范围运动过程中的柔顺性仍是一个需要解决的问题。本项目针对四足机器人的柔顺控制问题，在前期研究的基础上，借助虚拟模型控制方式、加速度规划方式、稳定裕度分析、深度学习算法等方式，对四足机器人足地交互过程中的柔顺性和稳定性进行研究，并将研究思路及成果推广至以机器人为代表的的自动控制及人工智能领域，从而提高机器人在非结构环境中的适应性，进而推广智能算法的研究工作。