Adaptive admittance control of 6-DoF parallel robot with sensorless force observer and collision avoidance

Parallel robot with human-robot collaboration (HRC) functionality to realize complex assembling has attracted much attention recently. This paper presents an adaptive admittance control by a six degree-of-freedom (DoF) parallel robot for the HRC assembling of large and heavy parts. A novel friction identification scheme concerning velocity and loads is proposed. By combining model-based and model-less methods, LuGre friction model and support vector regression (SVR) learning algorithm are applied to identify the nonlinear frictions of the robot. Considering possible collision resulted from large inertia of the parts, contacting force between parts is isolated from interacting force of operator. A collision detection scheme is defined by monitoring and filtering frequency of external forces with a bandpass filter. To ensure safety, efficiency and accuracy of the HRC assembling of large and heavy parts, an adaptive admittance control is developed. Update law of damping coefficient, coping with collision avoidance, is proposed. Experiments are conducted. External force estimator is compared with a force sensor. Deviations between estimated and measured forces are less than 10%. Large contacting force can be found by the collision detection scheme. The velocity and force of HRC assembling with different load show satisfactory behavior.