Posted in | News | Consumer Robotics | Humanoids

AI-Controlled Exoskeletons Improve Locomotion and Restore Mobility

Researchers from the New Jersey Institute of Technology developed a technique to train robotic exoskeletons using AI and computer simulations. This method will help users conserve energy when running, walking, and climbing stairs. The study was published in the journal Nature.

Nature: Experiment-free Exoskeleton Assistance via Learning in Simulation - Main Video

Video Credit: New Jersey Institute of Technology

The technique used in the current study is not limited to the hip exoskeleton; it can be used for a broad range of assistive devices.

It can also apply to knee or ankle exoskeletons, or other multi-joint exoskeletons.

Xianlian Zhou, Associate Professor and Director, BioDynamics Lab, New Jersey Institute of Technology

According to Zhou, it can also be used for prosthetics below or above the knee, providing immediate benefits for millions of able-bodied and mobility-impaired individuals.

Our approach marks a significant advancement in wearable robotics, as our exoskeleton controller is exclusively developed through AI-driven simulations. Moreover, this controller seamlessly transitions to hardware without requiring further human subject testing, rendering it experiment-free.

Xianlian Zhou, Associate Professor and Director, BioDynamics Lab, New Jersey Institute of Technology

This discovery could help people who struggle to move around, such as older adults or stroke survivors, without requiring them to be present in a lab or clinical setting for prolonged testing. Ultimately, it opens the door to improving accessibility and regaining mobility for regular in-home or community living.

This work proposes and demonstrates a new method that uses physics-informed and data-driven reinforcement learning to control wearable robots to directly benefit humans.

Hao Su, Corresponding Author and Associate Professor, Department of Mechanical and Aerospace Engineering, North Carolina State University

Exoskeletons have the potential to enhance human motor function for a broad range of users, including those requiring long-term support due to impairments or injury recovery. However, long-term human testing and regulatory restrictions have prevented their widespread use.

The research aimed to enhance the autonomous control of embodied AI systems, where an artificial intelligence program is incorporated into a tangible technology.

This work builds on earlier reinforcement learning-based research for lower limb rehabilitation exoskeletons, also a joint effort between Zhou, Su, and several others. It focuses on teaching robotic exoskeletons how to help able-bodied persons with a range of movements.

Previous achievements in reinforcement learning have tended to focus primarily on simulation and board games. Our method provides a foundation for turnkey solutions in controller development for wearable robots.

Shuzhen Luo, Assistant Professor and Study First Author, Embry-Riddle Aeronautical University

Luo previously worked as a postdoc at both Zhou’s and Su’s labs.

Typically, users must invest several hours in "training" an exoskeleton to ensure the technology accurately determines the necessary force. This training also helps the exoskeleton learn the precise timing for applying that force to aid users in walking, running, or climbing stairs.

The new method enables users to immediately use the exoskeletons, as the closed-loop simulation integrates exoskeleton controllers with physics models of musculoskeletal dynamics, human-robot interaction, and muscle reactions. This approach generates efficient and realistic data, iteratively improving control policy through simulation.

The device is pre-configured to function immediately out of the box. If researchers make hardware advancements in the lab through increased simulations, upgrading the controller on the device is also feasible. Prospects for this initiative in the future include creating personalized, specially designed controllers that help users with a range of everyday tasks.

Su said, “This work is essentially making science fiction reality—allowing people to burn less energy while conducting a variety of tasks.”

For instance, according to tests conducted on human subjects, when walking in the robotic exoskeleton as opposed to walking outside of it, study participants consumed 24.3 % less metabolic energy. When running in the exoskeleton, participants used 13.1 % less energy; when climbing stairs, they used 15.4 % less energy.

Although the new approach focuses on helping persons with mobility limitations use assistive technology, the study concentrated on the researchers' work with able-bodied individuals.

Su said, “Our framework may offer a generalizable and scalable strategy for the rapid development and widespread adoption of a variety of assistive robots for both able-bodied and mobility-impaired individuals.”

We are in the early stages of testing the new method’s performance in robotic exoskeletons being used by older adults and people with neurological conditions, such as cerebral palsy. And we are also interested in exploring how the method could be used to improve the performance of robotic prosthetic devices,” concludes Su.

This study was funded by the National Science Foundation, the National Institute on Disability, Independent Living, and Rehabilitation Research, the Administration for Community Living’s Switzer Research Fellowship Program, and the National Institute of Health.

Journal Reference:

Luo, S., et al. (2024) Experiment-free exoskeleton assistance via learning in simulation. Nature.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this news story?

Leave your feedback
Your comment type

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.

Read the full Terms & Conditions.