Scientists have made a framework for robots that builds safe routes through unstructured terrain based only on the movements of a human operator.

Existing approaches to route planning for robots fall into two categories:

The first type is to entrust the control of the robot to a trained person: he completely controls its movement and trajectory
The second type is to teach robots to plan their path and move autonomously.
Until now, the second type of robot control remains unreliable, since it is not always possible to predict the terrain and random events along the way. Also in this case, expensive equipment and sensors are usually required.

In a new work, researchers at the University of Trento have developed an alternative structure for planning the trajectory of a robot. It allows robots to determine and learn safe paths to their destination based on the movements of a human operator.

Safety is paramount if a person interacts with a robot in an unstructured space full of other people. The main goal of our work was to delegate the routine of route planning from robot to human, who, however, should focus only on which route to take. The robot is able to remember the path traveled and use it in future missions.

Alessandro Antonucci, one of the researchers

The approach developed by Antonucci and his colleagues greatly simplifies the trajectory planning task and does not require expensive sensors or other high-tech software components. Basically, the framework allows robots to recognize a human operator in order to capture his actions and movements.

The robot can distinguish between people and determine the distance to objects.