Adaptive Control of an Electrically Driven Exoskeleton Robot (Theory and Experiments)

Purpose: This paper presents a new adaptive tracking controller combined with a perturbation observer for controlling a rehabilitation robot during passive assistance tasks. The control strategy tries to mimic the robot's dynamics model and provides effective compensation for the robot's non-smooth, nonlinear limits, which are caused by factors such as backlash, hysteresis, deadzone, and saturation in its actuators. Methods: The main difference between the proposed approach and the conventional function approximation technique (FAT) is the elimination of the usage of a basis function to estimate the dynamic model law and the acceleration feedback. Furthermore, since the robot will be worn by humans, the load of human upper limbs may exert some resistant dynamics that will be regarded as external forces, resulting in degraded robot performance. To resolve the fact that the output of a newly designed disturbance observer is directly connected to the existing loop, resulting in a robust and active control system that removes severe nonlinear restrictions. Results: The proposed solution eliminates the need for knowledge of any of the exoskeleton robot's dynamic properties, including actuator dynamics and external forces, to achieve the desired performance. This study also includes simulation findings and comparison investigations to validate the benefits of the proposed approach. Conclusion: The results of experiments and simulations are very similar, which shows that the proposed approach proves its efficiency in real-time application.