by Yohan J. John, Harvard John A. Paulson School of Engineering and Applied Sciences

The Slade Lab applies insights gleaned from AI-generated simulations to build real-world biomedical devices with the potential to improve the everyday lives of people with motor challenges. Left, a member of the Slade Lab uses a prototype of an "exoskeleton" to assist people with walking difficulties. Credit: The Slade Lab

Many of the skills that humans consider intelligent involve conscious effort, including prediction, long-term planning, and abstract problem solving. Movement, by contrast, is something most people take for granted. It seems effortless and intuitive—a world away from the kinds of activities typically evoked by the word "intelligence."

But for Patrick Slade, Kempner Institute associate faculty member and an assistant professor of bioengineering at Harvard's School of Engineering and Applied Sciences (SEAS), skilled movement is one of the hallmarks of truly intelligent behavior, and a major frontier for cutting-edge research. The Slade Lab uses bioengineering and artificial intelligence techniques to better understand neuromotor principles and develop technologies to help people move.

"There are many mobility disorders that disrupt our ability to control our bodies, " says Slade. "So we need to better understand neuromotor control in order to help rehabilitate people."

Intelligence in motion

Neuromotor control refers to how neurons in the brain and the rest of the nervous system engage with muscles to produce voluntary movements. Slade explains that the brain is constantly assessing and balancing different kinds of motor needs as the body moves.

"If you're late for the bus you're going to prioritize speed, and you might not care if you're sweaty, or using a lot of energy, " he says. "But if it's icy outside, you may prioritize stability, and that's going to change how you move. In other situations, like hiking, you might need to be very careful where you're placing your feet."

Humans and other animals manage to juggle these parallel trade-offs—speed, energy, stability and so on—in real time, while pursuing various goals such as catching the bus or reaching the summit of a mountain. The key question that is common to motor neuroscientists and bioengineers is this: how does the brain manage to pull off this kind of "intelligence in motion"?

Using AI to better understand neuromotor control

Slade's current research project with the Kempner Institute involves trying to better understand the basic scientific principles and mechanisms underlying intelligence in motion by integrating neuroscientific, machine learning, and biomechanical approaches.

At the Kempner, Slade's team is analyzing large datasets of measurements from nerves and muscles using state-of-the-art AI techniques. Using information gleaned by analyzing large datasets of real human movement, the Slade Lab is developing skeletomuscular simulations, which are computer simulations of how the muscles control the skeleton.

These virtual imitations of human movement offer an improved understanding of how the body moves. Slade and his team can apply the insights gleaned from AI-generated simulations to build real-world biomedical devices with the potential to improve the everyday lives of people with motor challenges.

The Slade Lab has developed a "self-driving" cane for people with limited vision. Of the cane, Slade says, "It's like the user has their own personalized self-driving car, but they get to walk themselves." Credit: The Slade Lab

Building practical tools for improving lives

Applying insights about neuromotor control, the Slade Lab has developed prototypes of exoskeletons and other prosthetic devices to assist people with motor disabilities. These devices, worn by the user, can gather neuromotor information from sensors and use AI to extract crucial movement-related properties, and interpret and adapt that information to assist with motor activity. For example, an exoskeleton worn like a harness over the legs might provide additional force if the user needs to make it up a hill.

Slade's current focus is on building assistive devices for people with limited vision. This includes a self-driving robotic cane to help with partial visual impairment, as well as a smartphone that's worn around the neck. Using principles of smart computer vision models, the phone can understand the environment and give the user real-time audio feedback to help them navigate.

Of the cane, Slade says, "It's like the user has their own personalized self-driving car, but they get to walk themselves." Beyond building general capabilities of these devices, Slade emphasizes the importance of personalizing the devices for each user's specific needs. "One rule of thumb we've seen in our studies is that personalizing physical assistive medicine approximately doubles the benefits of the device, " says Slade.

And making the experience of using an assistive device feel natural for the user is paramount. A tool can be clunky and uncomfortable, or it can feel like a part of your own body. "The whole crux of what we hope to do is make the human-robot interaction seamless to the user, " says Slade.

Sparking new research directions

For his part, Slade says that the Kempner Institute has offered an opportunity to expand his thinking about the science of motor control and has led to promising new ideas.

"At the Kempner, I get to interface with all these different scientists, and that's helped us better understand many aspects of the science of motor control, " says Slade. "It's definitely sparked a lot of new directions for our research."