Carrying a payload directly on the body is an unavoidable aspect of human life. Human bipedal locomotion knows no equal: people travel on foot to virtually every corner of the globe. Despite the efficiency and convenience of wheeled apparatus, uneven terrain, enclosed environments and accessibility limits require virtually every transportation task to include a phase in which material goods must be physically carried by a person. As of today, no artificial intelligence or programmed behavior has been able to match a human’s ability to balance and maneuver in unstructured real-world environments. The Berkeley Lower Extremity Exoskeleton solves the problem of supporting and carrying heavy loads on the body and allows a person to navigate unencumbered by the weight of the payload they are carrying. The Berkeley Lower Extremity Exoskeleton is an anthropomorphic and energetically autonomous robotic device comprised of two legs, a backpack, a harness system and a control computer that provides a wearable load support platform. This thesis presents a control scheme called Sensitivity Amplification Control that enables an exoskeleton to support a payload and shadow the movement of the wearer in an intuitive and unobtrusive manner. The control algorithm developed here increases the closed-loop system sensitivity to its wearer’s forces and torques without any measurement from the wearer. This strategy requires an accurate dynamic model of the system but does not require direct measurements from the human. The trade-off between not having sensors to measure human action and the sacrificed robustness due to model parameter variation is described. A modification to the controller is also explored that partially circumvents this limitation.

Категория: Научные статьи | 1 Комментарий »

Комментарии