The algorithms with which robots calculate the trajectory of their movement from point "A" to point "B" exist long ago. There is nothing surprising. However, the movement of the robot in place, with the animated movement of pedestrians, is a much more complicated task. After all, you need to move according to generally accepted rules and norms, trying not to get in the way of people. Engineers at the Massachusetts Institute of Technology developed the algorithm just for such cases and successfully tested it.
The robot, which participated in the field trials, was equipped with several RGB-cameras, as well as a scanning sensor LiDAR. The software provided for the recognition of faces of surrounding people to respond to their appearance in the field of view of the robot and to track the trajectories of moving pedestrians relative to its current location. The robot neatly paves the route, trying to go around static obstacles and do not interfere with the people ahead. Its average speed is 1.2 meters per second.
At the same time, the robot not only monitors obstacles in its path and calculates trajectories of pedestrian traffic, but also tries to adhere to social norms adopted in a particular country. In the United States, as in Russia, people are used to right-wing traffic. Therefore, the robot tries to cling to the right side of the openings and corridors that cross. The settings can be quickly changed to alternative ones, after which the robot will accept the movement on the left side as normal.
The American automobile company Ford undertook to sponsor the project MIT . The results of the research will be made public at the forthcoming conference of the Institute of Electrical and Electronics Engineers, dedicated to the development of robotics, to be held next month.