Soft robotics: Design and control concepts for robots interacting with humans
The talk will summarize main design and control ideas of actively controlled compliant systems using as examples the DLR arms, hands, and the humanoid manipulator Justin. Variable, controllable impedance (stiffness, damping, and mass) is used to optimally adapt the systems to the task and the operating environment. Nonlinear observers and compliant reaction strategies are designed for detecting robot failure or collisions with humans and the environment and thus assure safety of humans in any circumstances during the interaction with the robots.
To characterize these set of robot design and interaction features we coined the term “Soft Robotics”. The presented theoretical framework extends the classical impedance control approaches for considering compliance both in the mechanical actuation as well as in the controller and treat both in a unified way.
Inspired by the human musculo-skeletal system, antagonistic, variable compliance actuators are a next logical step towards robots approaching the performance of their biological counterparts. Their control is providing an exciting link to neuroscience by helping to understand human motion principles and validate biological sensory-motor control hypotheses.
The presented approaches are used in space, service, and medical applications, as well as in a new type of industrial robotics focused on skilful assembly and cooperation with humans.
The talk will include practical examples on the use of impedance in some of these fields.