Sebastian Doliwa and Muhammad Ayaz Hussain and Tim Sziburis and Ioannis Iossifidis
Timing plays a vital role in the generation of naturalistic behavior satisfying all constraints arising from interacting with a dynamic environment while adapting the planning and execution of action sequences on-line.
In biological systems, many of the physiological and anatomical functions follow a particular level of periodicity and stabilization. The main aspect thereof is stabilizing movement timing against limited perturbations.
In the current work, we develop a model for timed movement preserving the serial order of actions by maintaining an approximately constant overall movement time which guarantees a temporal stabilized behavior of the system.
Action sequences are generated by dynamical system attractors providing a framework for robust incorporation of fluctuating sensor information. Movement satisfying a specific timing is of rhythmic nature which enables us to model the control of timed motion through stable limit cycles utilizing a Hopf oscillator.
We demonstrate the overall system both in simulation and on a physical robotic system by means of intercepting a moving target in three-dimensional space. For the prediction and correction of the target and current positions, an extended Kalman filter (EKF) is implemented and a modified Hopf oscillator generates the velocity profiles for the robotic arm.
In both videos, it is shown that first the implemented Kalman filter estimates the position of the ball, which then provides the Hopf oscillator with the correct set of parameters to ultimately generate the velocity profile for the robot arm to hit the ball. Limiting factor for the demostration was that only a small amount of joint acceleration was possible at the kuka lwr, resulting in relatively slow arm movements..