Position control and compliance control have been of critical importance for the exoskeleton robots. However, the uncertain disturbances from both the internal and external notably deteriorate the performance, e.g., causing large control error and even instability of the interaction between the human and exoskeleton, which poses significant challenges for the design of robust controllers for the exoskeleton robots. For the conventional position controllers, some of them are capable of achieving fast convergence of tracking error in finite-time, however, the control signals suffer from severe chattering. For the conventional compliance controllers, pursuing the desired compliance behavior in an accurate way while still guaranteeing stable interaction under disturbances is still open to i...[ Read more ]

Position control and compliance control have been of critical importance for the exoskeleton robots. However, the uncertain disturbances from both the internal and external notably deteriorate the performance, e.g., causing large control error and even instability of the interaction between the human and exoskeleton, which poses significant challenges for the design of robust controllers for the exoskeleton robots. For the conventional position controllers, some of them are capable of achieving fast convergence of tracking error in finite-time, however, the control signals suffer from severe chattering. For the conventional compliance controllers, pursuing the desired compliance behavior in an accurate way while still guaranteeing stable interaction under disturbances is still open to investigate.

In this thesis, we focus on the robust position and compliance control for the exoskeleton robots. The aims lie in three aspects. Firstly, we want to achieve finite-time convergence of error with smooth control action for the robust position control under disturbances. Secondly, we intend to investigate a complementary control framework, which decouples the design of admittance performance and robustness without potential tradeoff. Thirdly, we hope that the proposed robust position and compliance control methods are able to achieve satisfactory performance on an exoskeleton robot.

For the robust position control, we develop a chattering-free sliding mode controller enhanced by a disturbance observer. We show that the proposed method has finite-time convergence and smooth control signal even when there exists a large disturbance with bounded derivative. To verify the performance, simulations and experiments are conducted, which show that the proposed method achieves superior position tracking performance compared with the commonly-used methods.

For the compliance control, we propose a complementary admittance control framework so that the admittance performance and robustness can be regulated separately. We show that perfect admittance control is achievable by a full-information feedback controller, where the actual admittance exactly matches with the desired model and is guaranteed with passivity. The robustness is modulated by a robust regulator, which takes effect only when there exist disturbances. Algorithms for parameterizing the controller gains are developed. Both simulations and experiments are presented to show the efficacy of the proposed method in comparison with other methods.

Finally, we apply the robust position and admittance control methods to the lower limb exoskeleton, and propose a hybrid control framework. The complementary admittance controller is used to reduce the interaction forces/torques for the swing leg and the knee joint of the stance leg, while the robust position controller is utilized in the hip joint of the stance leg to support the upper body. Experiments with subject wearing the exoskeleton are presented to show its validity.