Effects of Assistance Delay on Joint Mechanics and Energetics in Biological Torque Control of a Hip Exoskeleton

Jimin An, Ryan Lee, Jingshu Peng, Eni Halilaj, and Inseung Kang

We tested whether assistance delay changes energetic and joint-level benefits across level-ground, ramp-ascent, and ramp-descent walking.

Assistance timing changed mechanical unloading, but not metabolic benefit, and no tested delay improved the evaluated outcomes during ramp descent.

Study-at-a-glance

8
Participants

3
Locomotor
tasks

5
Assistance
delays

4
Evaluated
outcomes

Biological torque estimates are converted into assistance through a fixed state-to-torque mapping

The controller receives IMU inputs and estimates the biological hip joint moment using a temporal convolutional network (TCN).
A predefined state-to-torque mapping function then scales, delays, and filters each estimate to generate the exoskeleton torque applied to the user.

We varied assistance delay across three biomechanically distinct walking tasks

Level ground, ramp ascent, and ramp descent
Five delays from 40 to 320 ms
Metabolic, joint-work, and subjective outcomes

Assistance delay affected mechanical unloading, but not net metabolic rate

Mechanical unloading: Shorter delays generally produced greater unloading.
Metabolic benefit: Net metabolic rate showed no detectable response to delay.
Ramp descent: No tested delay significantly improved the evaluated outcomes.

Subjective ratings vaired substaintially and showed no consistent group-level preference

Group response: Subjective ratings showed no detectable group-level response to delay.
Individual variation: Ratings varied substantially across participants.
Evaluation role: Subjective feedback should complement, rather than replace, metabolic and biomechanical outcomes.

Conclusion:
Mapping design should be guided by the target outcome and joint-level mechanics, not task labels alone.

Google Sites

Report abuse