摘要
针对四足机器人难以跨越大尺寸障碍物的问题,提出了一种仿生腾空跳跃的轨迹优化控制方法 .首先,对马里努阿犬的越障运动进行测试和仿生学分析,得到腾空跳跃的参考轨迹.其次,根据摩擦锥、轨迹平滑、运动学和动力学等约束条件,基于轨迹优化的方法对仿生参考轨迹进行离线优化,生成四足机器人的期望跳跃轨迹,得到跳跃所需的足端位置、关节力矩和关节角度序列.再次,设计分阶段的跳跃控制器,采用关节和质心PD(Proportional Derivative)的组合控制策略,实现对期望跳跃轨迹的跟踪,采用变刚度虚拟模型控制实现对落地姿态的调整和足底力的缓冲.最后,在Webots仿真环境中完成了1 m高度的自由落地和跳跃0.75 m高度桌子的仿真实验.研究结果表明:仿生轨迹优化与控制算法能够使四足机器人跳跃大尺寸障碍物并且具有平稳落地的效果,对于提高四足机器人的高性能运动能力以及拓展其应用环境具有参考价值.
A trajectory optimization control method inspired by biomimetic leaping is proposed to address the challenge of traversing large-scale obstacles for quadruped robots.Firstly,tests and bionic analysis of hurdle jumping in Malinois dogs are conducted to derive a reference trajectory for leaping.Subsequently,considering constraints such as friction cones,trajectory smoothing,kinematics,and dynamics,an offline optimization method is applied to the biomimetic reference trajectory to generate the desired jumping trajectory for quadruped robots.This process yields the necessary foot positions,joint torque,and joint angle sequences for executing the leap.Thirdly,a staged jumping controller is designed,employing a combination of joint and centroid Proportional Derivative(PD)control strategies to track the desired jumping trajectory.Additionally,variable stiffness virtual model control is employed to adjust the landing posture and cushion foot forces.Finally,simulation experiments involving free fall from a 1 meter height and leaping over a 0.75 meters high table are conducted using the Webots simulation environment.The research findings demonstrate that the biomimetic trajectory optimization and control algorithm enable quadruped robots to leap over large obstacles with smooth landings.This capability enhances their high-performance locomotion capabilities and broadens their potential application environments.
作者
张秀丽
赵浩宇
吴佳宁
许鹏
ZHANG Xiuli;ZHAO Haoyu;WU Jianing;XU Peng(School of Mechanical,Electronic and Control Engineering,Beijing Jiaotong University,Beijing 100044,China;China North Vehicle Research Institute,Beijing 100072,China)
出处
《北京交通大学学报》
CAS
CSCD
北大核心
2024年第3期161-170,共10页
JOURNAL OF BEIJING JIAOTONG UNIVERSITY
基金
国家重点研发计划(KMB05322001534)。
关键词
四足机器人
仿生学
轨迹优化
跳跃
虚拟模型控制
quadruped robots
bionics
trajectory optimization
jumping
virtual model control
