Closed-loop Control using Electrotactile Feedback Encoded in Frequency and Pulse Width

Abstract

Sensory substitution by electrotactile stimulation has been widely investigated for improving the functionality of human-machine interfaces. Few studies, however, have objectively compared different ways in which such systems can be implemented. In this study, we compare encoding of a feedback variable in stimulation pulse width or stimulation frequency during a closed-loop control task. Specifically, participants were asked to track a predefined pseudorandom trajectory using a joystick with electrotactile feedback as the only indication of the tracking error. Each participant performed eight 90 s trials per encoding scheme. Tracking performance using frequency modulation enabled lower tracking error (RMSE: Frequency modulation: 0.27±0.03; Pulse width modulation: 0.31±0.05; p<0.05) and a higher correlation with the target trajectory (Frequency modulation: 83.4±4.1%; Pulse width modulation: 79.8±5.2%; p<0.05). There was no significant improvement in performance over the eight trials. Furthermore, frequency-domain analysis revealed that frequency modulation was characterized with a higher gain at lower error frequencies . In summary, the results indicate that encoding of feedback variables in the frequency of pulses enables better control than pulse width modulation in closed-loop dynamic tasks.

Task-Dependent Adaptations in Closed-Loop Motor Control Based on Electrotactile Feedback

“…Closed-loop control with electrotactile stimulation was assessed in several studies [19]- [24] but they focused on characterizing performance without estimating the model of the human controller. To the best of our knowledge, just two studies reported the human transfer functions when using electrotactile feedback; however, only for position control mode, showing that the transfer functions resembled low-pass filters [25], [26]. The findings for tactile feedback are so far in agreement with the theory of human manual control, but as described above, the assessment was overall scarce.…”

“…We assessed the performance and estimated the model of the human controller in both control modes (position versus velocity) when the feedback was delivered using encoding in stimulation frequency. While most previous studies employed amplitude modulation (e.g., [19]- [22]), frequency modulation was selected since we have shown that it enables superior performance in the same type of closed-loop control task as used in this study [26]. Our a priori expectation was that a human controller adapts to different plant dynamics using electrotactile feedback in the same way it adapts when the feedback is transmitted visually.…”

Statistical Characterization of Closed-Loop Latency at the Mobile Edge

“…This experiment demonstrated that the average latency increases significantly with the network load. A closed-loop compensatory tracking task is performed in the closed-loop control using tactile input in [25], where the feedback is encoded to the user using frequency and amplitude modulation schemes. Significant time delay on the order of several hundred milliseconds has been noticed in this experiment.…”

“…where T r denotes the time elapsed in compressing the raw sensing data with compression ratio Q (see ( 19)), T f tx denotes the time elapsed in transmitting the compressed data (see ( 10)), T f d denotes the delay incurred in decompressing the compressed sensing data (see (25)), and T f c denotes the time elapsed in processing the sensing data (see ( 17)). T pf tx is the time elapsed in transmitting the low-level command (having number of packets N p f ) extracted from sensing data to the HO (see (10)).…”

2020 Index IEEE Transactions on Haptics Vol. 13

“…Closed-loop Control using Electrotactile Feedback Encoded in Frequency and Pulse Width. Dideriksen, J.L., +, TOH Oct.-Dec. 2020 Xu, H., +, TOH July-Sept. 2020 483-492…”