Autonomy Talks - 01/02/2021
Speaker: Matthias Hofer, Institute for Dynamic Systems and Control, ETH Zürich
Title: Development of an Inflatable Robotic Arm: Design, Learning Control and Vision-based Sensing
Abstract: Soft robots show promise to overcome challenges encountered with traditional rigid robots due to the versatility of the soft materials employed. Their intrinsic mechanical properties, such as their low weight and adjustable compliance are beneficial in terms of safety, allowing for close human-robot collaboration. In this talk, I will present the development of an inflatable fabric-based robotic arm which is pneumatically actuated. The capability to adjust the joint stiffness for different actuator configurations is discussed, including its implication on the dynamics of the robotic arm. The use of model-based learning control approaches such as Iterative Learning Control and offset-free Model Predictive Control are demonstrated to improve tracking performance and thereby enable e.g. a pick and place application and the catching of a ball with the soft robotic arm. Finally, a vision-based sensing approach is discussed which relies on cameras integrated into the soft system. Machine learning techniques are employed to map the camera images to the positional degrees of freedom of the robotic arm.
If you want to know more about Autonomy Talks, check [ Ссылка ]
