Model-based control and object contact detection for a fluidic actuated robotic hand

The necessity of position and force control for robotic systems is widely accepted. The high level control of an artificial hand has to base on a funded knowledge of the systems behavior and a low level control. This article presents a joint controller for an anthropomorphic robot hand driven by flexible fluidic actuators. These flexible and compact actuators are integrated directly into the finger joints, they can be driven either pneumatically or hydraulically. A both theoretically and experimentally obtained model of the fluidic driven finger joints is described. Based on this nonlinear model incorporating the viscoelastic behavior of the actuators material, an adaptive position controller is presented. The actuator characteristics allow us to enhance the position controller with a model based torque limitation. Thus an additional force sensor is not necessary. Likewise, based on the actuator model, an object contact detection is introduced. The functionality of the addressed controller features is shown by experimental results.