Search papers, labs, and topics across Lattice.
This paper introduces a machine learning-based perception module for exosuits that estimates three walking modes (ascending/descending stairs, level ground) using only inertial data from two sensors. The approach forecasts both future and past timesteps, enabling control and self-labeling for online model adaptation to new users. Experiments on real-life datasets demonstrate the effectiveness of the user-tailored perception module, validated through closed-loop integration with an exosuit in a single-subject online experiment.
Exosuits can now adapt to individual users in real-time by forecasting walking modes from just two inertial sensors, paving the way for more personalized and effective assistive robotic devices.
Assistive robotic devices, like soft lower-limb exoskeletons or exosuits, are widely spreading with the promise of helping people in everyday life. To make such systems adaptive to the variety of users wearing them, it is desirable to endow exosuits with advanced perception systems. However, exosuits have little sensory equipment because they need to be light and easy to wear. This paper presents a perception module based on machine learning that aims at estimating 3 walking modes (i.e., ascending or descending stairs and walking on level ground) of users wearing an exosuit. We tackle this perception problem using only inertial data from two sensors. Our approach provides an estimate for both future and past timesteps that supports control and enables a self-labeling procedure for online model adaptation. Indeed, we show that our estimate can label data acquired online and refine the model for new users. A thorough analysis carried out on real-life datasets shows the effectiveness of our user-tailored perception module. Finally, we integrate our system with the exosuit in a closed-loop controller, validating its performance in an online single-subject experiment.
