How AI Is Personalizing Wearable Robotics and Assistive Devices

Summary

Patrick Slade, a biomedical engineering professor at Harvard SEAS, explains how artificial intelligence and machine learning can personalize wearable robotics and assistive devices. These systems can tailor force, timing, and treatment goals for individuals recovering from stroke, managing knee pain, or navigating with visual impairments.

Why it matters

Assistive technologies have long faced a one-size-fits-all approach. A fixed controller that benefits one person may negatively impact another. Slade’s work argues that combining biomechanical sensors with AI can bridge this gap, transforming devices from passive data collectors into truly adaptive assistants. This is critical for the aging population and individuals with disabilities who rely on these tools for daily mobility and independence, as the risks of a poor fit are significant.

Highlights

• Consumer smartwatches can misestimate energy expenditure by 40% to 80%. This is largely because most of these devices rely on indirect signals like wrist movement rather than the leg muscle activity that actually determines calorie cost.
• Slade’s lab combines smartphone accelerometer data with physics-based machine learning models. This approach provides more accurate estimates of joint load, gait asymmetry, and energy cost compared to off-the-shelf wearables.
• A system is being developed to help visually impaired or low-vision users navigate curbs, crosswalks, and obstacles in real time, utilizing a smartphone camera, GPS, and an open-ear audio system.

Critical take

The text reads more like a research vision presented in a Q&A format than a results report. It lacks comparative user trial data, error modes, or performance benchmarks. Furthermore, the claim that self-adapting consumer wearables will hit the market 'within the next few years' is an unsubstantiated prediction, lacking supporting evidence or a regulatory roadmap.