Optimal Velocity Control of Spherical Robots Based on Offset-free Linear Model Predictive Control

Since the sensors of spherical robots rotate with the longitudinal axis, an efficient velocity controller must also account for robot's attitude. Therefore, this paper proposed an optimal velocity controller for spherical robots based on the offset-free linear model predictive control (LMPC), which controls the velocity while considering robot's attitude, motor's current and other variables. The prediction model of LMPC was generated by linearizing and discretizing the dynamic model and introducing the disturbance model. Then the kalman filter (KF) was used to estimate the states and disturbance. After that, combined with LMPC, the offset-free tracking of different velocities under different terrains was achieved by fully considering velocity, attitude and current in the target function. Finally, a series of physical experiments show that the proposed controller ensures fast convergence, strong stability and no bias, while the velocity changes more smoothly, the attitude changes less and the current output is more stable.