摘要
针对双足机器人变步长步行需求,提出一种基于变质心高度策略的变步长步态生成方法。首先,通过理论计算结合样机实验,得出相同步行周期下不同步长对应的机器人能耗最低的质心高度,作为变高度过程中质心高度的参考值;其次,基于模型预测控制(MPC)建立质心状态方程,构建综合质心位轨迹、速度优化和落脚点控制的优化目标函数,考虑零力矩点(ZMP)稳定裕度约束,建立二次约束二次规划模型(QCQP),生成最优质心轨迹;最后,以NAO机器人为实验平台进行变步长步行对比实验。研究结果表明:采用模型预测控制方法,机器人步行过程中ZMP始终保持在稳定范围内,实现了稳定步行;基于变质心高度的步态生成策略,根据步长变化调整机器人质心高度,能适应0.04~0.10 m变步长步行,相比于定高度线性倒立摆模型具有更强的步长调节能力;针对所设计的变步长序列,基于变质心高度的变步长方法相比于传统恒定质心高度方法平均能耗降低约16%。
To deal with the biped walking with variable step-length of humanoid robot,a gaits generation method was proposed based on variable center-of-mass(COM)height strategy to achieve variable step-length walking.Firstly,the energy efficient COM height corresponding to the step-length was carried out through the theoretical calculation combined with experiments and then the COM height was adopted as the reference value.Then,the COM state model was established based on model predictive control(MPC),and the cost function was proposed,where the COM trajectory generation,velocity optimization and footstep locations adjustment were incorporated.Considering the zero moment point(ZMP)stability constraints,a quadratically constrained quadratic program(QCQP)was modeled to generate optimal COM reference trajectory.Finally,the variable step-length walking experiment was carried out on the NAO robot.The results show that the ZMP strictly keeps in the constrained area so that the robot can achieve stable biped walking based on MPC.The strategy based on variable COM height can adjust COM height according to step-length and can adapt to the range of 0.04—0.10 m,resulting in stronger steplength adjustment ability.The energy of the variable step-length warking gaits based on proposed strategy reduces by 16%compared to traditional method based on 3D LIPM at constant height.
作者
周江琛
肖晓晖
ZHOU Jiangchen;XIAO Xiaohui(School of Power and Mechanical Engineering,Wuhan University,Wuhan 430072,China)
出处
《中南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2020年第4期944-952,共9页
Journal of Central South University:Science and Technology
基金
国家自然科学基金资助项目(51675385)。
关键词
变质心高度
变步长
模型预测控制
能量效率
双足步行
variable center-of-mass height
variable step-length
model predictive control
energy efficiency
bipedal walking
