Robotics in the workforce is our reality today.
However, far from zipping around like they do in popular sci-fi films, factory robots tend to move around shop floors inefficiently; they are often overly cautious to the point of freezing when they detect movement.
MIT believes they have found a solution to speedier and safer robots. Their new algorithm accurately predicts a person’s movement trajectory, aiding in collision avoidance and needless time-wasting.
RELATED: THESE AIRBAGS STOP ROBOTS FROM ACCIDENTALLY INJURING WORKERS
The project’s origins
In 2018, a team of MIT researchers collaborated with car manufacturer BMW in order to test new, safer ways in which humans and robots might work, at proximity, building car parts.
A test where humans crossed a replica factory floor, while a robot on rails glided across it, quickly outlined the main problems with existing algorithms.
The robot was programmed to stop momentarily if a person passed by. However, researchers found it would often freeze on the spot, overly cautious, long before anyone moved across its path.
Such accumulated, lengthy waiting times would slow down the manufacturing process considerably.
Speeding the process with the algorithm’s limitations, however, would have created a dangerous work environment.
The problem was traced to a limitation in the robot’s trajectory alignment algorithms (its motion predicting software). It could efficiently predict where a person was headed, but it struggled to know how long they would take.
A new trajectory alignment algorithm
The new system, created by the same team, aligns parts of a person’s trajectory with a database of reference movements. It learns from people’s movements and uses smart predictions to know how long a person will take to get from point A to point B.
For example, it knows a person who has just started moving is not likely to change direction immediately.
Simulations showed that robots using the algorithm were much less likely to freeze and would be able to more efficiently carry out their functions. Ultimately, it allows safer robots at work, without compromising their efficiency.
While there’s still a way to go before large, heavy, AI-controlled robots are deemed safe enough to maneuver freely around human workers, this is a big step in helping robots and AI to understand human behavior.
Source: Interesting Engineering