Nidhi Seethapathi

Even though robots can have motors that outlast our muscles and circuitry faster than our neurons, we humans are still better at walking. And we’re better at moving through new environments. The reason we’re so good at these activities has remained a mystery, in part because computer models used to understand human movement are split into separate components for feedback control and learning, and energy efficiency.

Nidhi Seethapathi, 34, has addressed this riddle by creating a unified model that can accurately predict how we move through our daily lives, learn new tasks, and navigate novel environments.  

“The field has two dominating ideas” for how a person prioritizes her next move, says Seethapathi. “One is minimizing the metabolic energy cost of moving. The other is moving in a way that is stable, safe, and error-free.” 

Seethapathi’s goal was to figure out how these sometimes competing constraints interact to keep us on our feet. She combined data sets drawn from natural movement in the real world with video analysis of people walking on a treadmill (both unencumbered and while being occasionally yanked by a bungee cord), and she gathered sensor measurements. Then she analyzed the results against how much energy our bodies consume when we run vs. walk, or when we adjust our gaits for different surfaces or in response to small errors in muscle activation. 

The result is a new computer model for human movement that Seethapathi hopes will lead to better exoskeleton suits, split-belt treadmills for stroke rehabilitation, and prosthetics. It could also inspire new exercises tuned for specific sports or medical conditions.