Lower limb prostheses that are self-powered and help users walk fairly naturally don’t work well on varying surfaces. Walking on paved concrete is very different from traversing a grassy field, but without knowing the terrain below it is impossible for a prosthetic to adjust accordingly. Now, researchers at North Carolina State University have developed a camera-based artificial intelligence system that can recognize a variety of terrains and signal a prosthetic to change how it moves to match the surface underneath.
The technology involves attaching a small camera near the prosthetic so it can see the ground ahead. Video is fed into a computer that was taught to identify different terrains, including grass, brick, concrete, and even tile. The device can also recognize whether there are staircase steps ahead and if they’re pointing up or down.
“Lower-limb robotic prosthetics need to execute different behaviors based on the terrain users are walking on,” said Edgar Lobaton, co-author of the study presenting this technology in IEEE Transactions on Automation Science and Engineering. “The framework we’ve created allows the AI in robotic prostheses to predict the type of terrain users will be stepping on, quantify the uncertainties associated with that prediction, and then incorporate that uncertainty into its decision-making. We came up with a better way to teach deep-learning systems how to evaluate and quantify uncertainty in a way that allows the system to incorporate uncertainty into its decision making. This is certainly relevant for robotic prosthetics, but our work here could be applied to any type of deep-learning system.”
“If the degree of uncertainty is too high, the AI isn’t forced to make a questionable decision – it could instead notify the user that it doesn’t have enough confidence in its prediction to act, or it could default to a ‘safe’ mode,” added Boxuan Zhong, lead author of the study.
Study in IEEE Transactions on Automation Science and Engineering: Environmental Context Prediction for Lower Limb Prostheses With Uncertainty Quantification