1D31 Smart-servo control by rigid body physics based methods(The 12th International Conference on Motion and Vibration Control)

DOI Web Site

Abstract

Smart-servomotors have been used in biomimetic robots due to their integrated sensor and controller design, however their control is made difficult due to difficult to simulate non-linearities. Control of smart-servos can be improved by proper simulation of motor dynamics dependent on non-trivial external forces. Although simulation has been successful to varying degrees in the past, a new method using rigid body physics which can be run several times faster than real-time is proposed. Motor torque and friction are modeled separately as targets and constraints on the system rather than calculated from motor state equations and directly applied, simplifying the motor model. The proposed method assumes the internal motor dynamics are negligible compared to the controller operations and friction components and can therefore be run at lower frequency without significant losses in accuracy. Simulation and control were verified by applying a sinusoidal target angle function to a three motor planar revolute arm with external loading applied at the end effector. The error between experimental and simulation results depended on the torque at the motor, with a minimum RMS error of 0.0140 radians at the effector motor when a 200g mass was applied to the arm and a maximum RMS error of 0.0259 radians at the base motor. Control of the simulation using experimental data with no external load showed an error of 0.0219 radians at the base motor. The results show that both direct simulation and online control have low error despite significantly simplified modeling.

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top