Wheel Slippage Compensation in Mobile Manipulators Through Combined Kinematic, Dynamic, and Sliding Mode Control
Moharam Habibnejad Korayem,Siavash Fathollahi Dehkordi,Narges Ghobadi
DOI: https://doi.org/10.1007/s13369-024-08718-y
IF: 2.807
2024-02-26
Arabian Journal for Science and Engineering
Abstract:In this article, an updated dynamic model of mobile manipulators is presented, which incorporates the effects of wheel slipping and skidding in the mobile base motion equations. This modified model is then used to design an improved control algorithm. The interaction between the robot's mobile platform and manipulator often leads to inaccuracies when using traditional control methods. To address this, a modified control strategy is proposed. Unlike previous research, which faced difficulties directly measuring wheel slippage and traction forces, this work implements a disturbance observer to estimate these unknown parameters. The control algorithm is then designed using the feedback from the observer estimates. First, kinematic control is used to guide the robot along the desired trajectory. Next, dynamic control augmented with the disturbance observer enables robust tracking. Specifically, a disturbance observer-based sliding mode controller is developed for dynamic control of the system. This is further optimized using the state-dependent Riccati equation method. Lyapunov analysis proves system stability and guarantees disturbance estimation errors converge to zero. Simulations on a model of a mobile manipulator confirm the effectiveness of the proposed method. When combining the disturbance observer and sliding mode control, the consumed torque of the wheels and arms is reduced on average by 0.42 Nm and 0.04 Nm, respectively. Defining a cost function and optimizing the torques with the optimal sliding mode control approach further decreases the required torque to 5.17 Nm compared to the basic SMC, while also reducing tracking error by 4.9 mm. Despite platform slippage, the controller performance keeps end effector errors small and within allowable bounds. Experimental validation on a Scout robot demonstrates the feasibility of implementing this method on physical systems. The robot is able to track desired trajectories with acceptable errors. The tracking error in experiments is approximately 54 mm, compared to 13 mm in simulation.
multidisciplinary sciences