Robots in industrial settings have become really good at performing highly repetitive actions at lightning speed. But when required to work alongside humans, they can be like teenage boys at a dance: all left feet, tripping on their partners, and freezing up after making a mistake.
Julie Shah wants to create a new generation of assembly line robots that will be more attentive and better coordinated with their human workmates. An associate professor of aeronautics and astronautics, and director of MIT’s Computer Science and Artificial Intelligence Laboratory’s (CSAIL) Interactive Robotics Group, Shah is designing factory-based machines that will better anticipate human movement, and sidestep rather than shut down to avoid injuring flesh-and-blood partners.
“We’re developing robot planning, decision making, and control technology that allows robots to adapt their actions, using a model of what a person is doing, or might do,” she says.
This vision of a safer and more efficient factory floor is already becoming a reality, through a series of collaborations with major corporations. At a Boeing factory that assembles commercial airplane fuselages, Shah has developed an algorithm that enables robots to continue working if nearby humans leave and return to the line. With the firm ABB, she has developed predictive models so robot systems can “track human arms for elbow-to-elbow work.”
Shah’s designs for robots and their industrial environments are intended to speed up production from start to finish. “Assembly lines are limited by the slowest station,” she says. Her hope is to staff each station, even those requiring the most dexterous work, with inherently human-safe machine assistants.
A recent breakthrough with partner BMW brings this future closer. In an industrial machine first, Shah successfully tested a freely mobile robot worker that moves on and off an automo- tive final assembly line with tools and materials, navigating safely, and gracefully, around human partners.