MIT specialists have developed an algorithm that will help robots guess where people are going. MIT researchers have found that robots sometimes make mistakes and freeze in place, whereas human crosses the road. The robot was programmed to stop momentarily if a person passed by
Automation is becoming more and more real, and this means that robots work together with humans. But there is a problem; robots often lousy predict where people are going, which leads them either to sinking or to the risk of colliding with their living colleagues.
Fortunately, MIT researchers have developed an algorithm that better predicts the ways of nearby people. Instead of just relying on the distance of points on the human body, like ordinary systems, the new approach aligns the segments of the human trajectory with a set of reference movements. Moreover, it also takes time into account – it knows that you are not going to change course if you just started to move.
As a result, robots must be more confident, but not reckless. In simulations based on earlier data on human movement, automata were less prone to panic and quickly returned to work after a person passed by.
There are many things to do before this algorithm can achieve the real world, but there is an obvious utility here. Factory and warehouse robots can work closely with people without slowing down or risking injury. MIT added that it can also be used for other interactions, such as recognizing actions and gestures. Robots could better understand human behavior in general.
It does not determine the distance between the robot and the points on the human body but is engaged in building a safe path to the path area.
Using this method allows you to improve the performance of the robot. So, devices, when people walk past them, usually freeze in place so as not to hurt them and not cause injury. The algorithm helps the mechanism to build a safer trajectory of movement, which not only does not harm the person but also allows you to not interrupt work.