Researchers develop an algorithm that tells robots where humans are going: Tool can soon allow humans and robots to work in a factory setting… or someday hunt us like prey

Monday, February 3, 2020
By Paul Martin

by: Lance D Johnson
Monday, February 03, 2020

Researchers from the Massachusetts Institute of Technology (MIT) have been working alongside auto manufacturing company BMW to develop an algorithm that tells robots how to predict human movement and how to respond accordingly.

Factory workers are beginning to work in close proximity to robots. Researchers are trying to enhance the efficiency of robots so that they can work alongside humans and judge their movements with precision. By anticipating human movements, robots will be able to complete tasks faster.

Algorithm allows robots to predict human movements, beneficial for factory settings

In factory tests, robots were tasked with delivering parts between work stations. Occasionally a human worker crossed the robot’s path. The robot, programmed to stop if a person passes by, would often freeze too soon, long before a person crossed its path. When a new trajectory alignment algorithm was put in place, the robot became braver and its motion predicting software became more precise. With the upgraded algorithm, the robots could accurately predict where a person was headed and could anticipate how long the person would remain in the projected path. The robot began to accurately predict how long it takes a person to stop and when they might double back across the robot’s intended path.

The MIT researchers found a way to accurately align partial trajectories in real time. The motion detectors not only sensed movement but they could accurately predict the timing of a person’s movement. When the enhancements were applied on the BMW factory floor, the robots gained tremendous judgment and skill, rolling through their movements without hesitation or stall. When human workers doubled back across a robot’s path, the robot had already predicted the movement and was already safely out of the way and still performing tasks.

“This algorithm builds in components that help a robot understand and monitor stops and overlaps in movement, which are a core part of human motion,” says Julie Shah, associate professor of aeronautics and astronautics at MIT. “This technique is one of the many ways we’re working on robots better understanding people.”

The motion prediction can also be used as a pre-processing step for other interactions between humans and robots such as action recognition and gesture detection. The algorithms will enable robots to recognize human movements and human behaviors, allowing robots to respond and make critical judgments about human interactions.

An algorithm like this could, one day, allow robots to hunt us down like prey

The Rest…HERE

Comments are closed.

Join the revolution in 2018. Revolution Radio is 100% volunteer ran. Any contributions are greatly appreciated. God bless!

Follow us on Twitter