van der Kooij Herman, Jacobs Ron, Koopman Bart, van der Helm Frans
Institute of Biomedical Technology, University of Twente, PO Box 217, 7500 AE Enschede, The Netherlands.
Biol Cybern. 2003 Jan;88(1):46-59. doi: 10.1007/s00422-002-0330-5.
Based on mechanical analysis, three gait descriptors are found which should be controlled to generate cyclic gait of a seven-link humanoid biped in the sagittal plane: (i) step length, (ii) step time, and (iii) the velocity of the center of mass (CoM) at push off. Two of these three gait descriptors can be chosen independently, since the CoM moves almost ballistically during the swing phase. These gait descriptors are formulated as end-point conditions and are regulated by a model predictive controller. In addition, continuous controls at the trunk and knees are implemented to maintain the trunk upright and to ensure weight bearing. The model predictive controller is realized by quadratic dynamic matrix control, which offers the possibility of including constraints that are exposed by the environment and the biped itself. Specifying step length and CoM velocity at push off, the controller generates a symmetric and stable gait. The proposed control scheme serves as a general-purpose solution for the generation of a bipedal gait. The proposed model contains fewer parameters than other models, and they are all directly related to determinants of bipedal gait: step length, trunk orientation, step time, walking velocity, and weight bearing. The proposed control objectives and the model of humanoid bipedal walking have potential applications in robotics and rehabilitation engineering.
基于力学分析，发现了三个步态描述符，在矢状面中生成七连杆类人双足机器人的循环步态时应控制这三个描述符：（i）步长，（ii）步时，以及（iii）蹬离时质心（CoM）的速度。这三个步态描述符中的两个可以独立选择，因为质心在摆动阶段几乎做弹道运动。这些步态描述符被制定为端点条件，并由模型预测控制器进行调节。此外，还对躯干和膝盖进行连续控制，以保持躯干直立并确保负重。模型预测控制器通过二次动态矩阵控制实现，这提供了纳入由环境和双足机器人自身所呈现的约束的可能性。指定蹬离时的步长和质心速度，控制器会生成对称且稳定的步态。所提出的控制方案作为双足步态生成的通用解决方案。所提出的模型比其他模型包含的参数更少，并且它们都与双足步态的决定因素直接相关：步长、躯干方向、步时、行走速度和负重。所提出的控制目标和类人双足行走模型在机器人技术和康复工程中具有潜在应用。
