Adaptive model-mediated control for enhanced force control and stiffness discrimination in teleoperation

Abstract

Model-mediated teleoperation control employs an environment model at the master side to compute feedback output to the master at a faster rate. This approach improves system stability in the presence of time delay. It has an advantage in task performance as it does not utilize adaptive dampers which can cause unpredictable force changes. Previous literature, however, reports experiment results showing that system stability cannot be guaranteed if the model is not accurate. This implies that model-mediated teleoperation does not generally perform well if the environment impedance is unknown in advance or varies during a task. This dissertation, first, analyzes the limitations of the previous model-mediated methods in terms of system stability and task performance. It is confirmed that the system in the presence of delay generates energy when the model stiffness increases with the pressed depth. Simulations and experiments show that the energy generation induces fluctuations in the master and destabilizes the system. This dissertation proposes a model-mediated control which guarantees system stability with time delay even if the environment model cannot be predefined accurately. The proposed method renders force feedback through a linear spring model similar to the previous methods. The reference point of the model, however, is fixed to the contact position between the slave and the environment, whereas the reference point in the proposed method is adaptively set responding to the model stiffness update. The reference point moves to a position where the output to the master does not change even when the model stiffness is updated. This makes the master’s state change only by the operator’s manipulation so that system stability is not affected by time delay. Simulations and experiments verify that the proposed method guarantees stability even in the presence of delay and model mismatch. User studies are conducted to quantitatively measure and compare task performance in the proposed method with the previous methods that improve the stability in model-mediated teleoperation. It is confirmed that the proposed method more accurately renders the environment force to the operator. It is also observed that the operator's force-controlling accuracy is not affected by time delay in the proposed method in contrast to other methods. The stiffness discrimination experiment confirms that subjects can discriminate up to 40.9% smaller differences in the presence of 100 ms time delay than when using other methods. Spearman correlation coefficient shows that the subjects’ JNDs are the least degraded by time delay in the proposed method.