Intelligent Walking of Humans and Robots

在该项目的第一阶段，研究表明，即使当行走在地下覆盖的不平坦处时，也会在人类行走时产生虚拟枢轴点(VPP)。考虑到目前实施的控制策略，在人形机器人萝拉中没有观察到这样的VPP。因此，该提案第二阶段的目标将是发现行走过程中出现VPP是否是一个受控或可控的目标，或者它是否是肌肉骨骼系统动态相互作用的内在原因。为了研究这一基本问题，一方面将对人类步态进行进一步的分析(A部分)，另一方面将进行研究，以发现如何为类人机器人建立相关的控制策略(B部分)。特别是，了解能够生成或修改VPP的可能的调谐参数将是很有意义的。因此，在A部分中，将创造行走条件，目的是消除VPP的外观(例如，通过倒退行走、踩高跷或在完成额外任务时)。此外，我们将测量神经系统疾病患者(帕金森和多发性硬化症)的步态，以分析这种情况如何改变和消灭VPP。在B部分，将对控制策略进行调整、修改(例如，在规划过程中放松对ZMP约束的限制)或重新定义，以便在人形机器人的步态过程中产生VPP。这些新的控制策略将首先在简单模型(如BSLIP模型)上进行研究，然后将在LOLA机器人的规划和控制中实施。

In the first phase of the project, it was shown that a Virtual Pivot Point (VPP) arises during human gait even when walking over a covered unevenness of the underground. Such a VPP is not observed in the humanoid robot Lola, given the current control strategies implemented. Therefore, the objective of the second phase of the proposal will be to discover if the appearance of a VPP during walking is a controlled or controllable goal or if it is intrinsic to the dynamic interplay in the musculoskeletal system. To investigate this fundamental question, further analysis of the human gait will be performed (Part A) on one hand and, on the other hand, research will be conducted to discover how a related control strategy can be built for humanoid robots (Part B). In particular, it will be interesting to understand the possible tuning parameters enabling to generate or modify the VPP. Thus, in Part A, walking conditions will be created with the aim to annihilate the appearance of a VPP (e.g by walking backwards, walking on stilts or while fulfilling additional tasks). Additionally, we will measure the gait of neurologically sick patients (Parkinson and multiple sclerosis) in order to analyze how such a condition modifies and annihilates the VPP. In Part B, the control strategy will be adapted, modified (for instance by relaxing the constraints on the ZMP-constraint during the planning) or redefined in order to generate a VPP during the gait of a humanoid robot. These new control strategies will be first investigated on simple models (such as the BSLIP model) and will then be implemented in the planning and control of the Lola robot.