Research on Interaction Force for the Human-robot System based on Double People Walking Experiments

Abstract

Walking requires metabolic energy, especially in load-carriage. Many wearable devices have been proposed to assist humans in walking with loads, such as elastic backpacks, exoskeletons, and supernumerary robotic legs (SRLs). Although remarkable progress has been obtained in this field, they still lack appropriate physical interaction control and response strategy, which may lead to crashing and dragging problems. Owing to the complexity of human walking, appropriate interaction force control is crucial to fulfilling the task of compliant human-robot cooperation. However, there is nearly no research on the human-robot interaction force during human walking presently. To address this issue, double people cooperative walking experiments have been designed to obtain the rules of timing and magnitude of the interaction force. Since human beings can adjust the step length and speed to different walking situations for energy saving, these two participants automatically enter into the cooperative walking pattern after a few steps at the beginning. The interaction force obtained from the cooperative walking pattern is likely to be energetically optimal for the overall system. The result shows that the double people can provide forward propulsion for the former in the double stance (DS) phase to reduce the metabolic cost by reducing the needed ground reaction force. It is also found that the data of the interaction force is positively correlated with the acceleration of the human center of mass (CoM). It also implies that in a human-robot system, the robot motion control system estimates the appropriate interaction forces by tracking the acceleration of the CoM, which may contribute to inspiring the design of the physical interaction control system of a human-robot system.