Researchers in robotics from the University of California at Davis have developed a control system which allows the robots to collect evidence suggesting that their leader is about to turn, predict where it will go and then follow him. “This is a fundamental problem in robotics,” says Sanjay Joshi, associate professor mechanical and aeronautical engineering at UC davis. Indeed, whether walking down the street, while driving on the highway or in many other situations, the man often collected deliberate signals and unconscious clues in order to predict what the others and act accordingly. The robots, however, more difficult to coordinate so, for example when the leader of a group turns a corner and disappears from the field of vision of its congeners. Studies in behavioral psychology have shown that a person is about to turn unconsciously a brief nod in the direction it is preparing to borrow so that others can follow.

A system inspired by studies of behavioral psychology
Humans use different signals and to build a predictive model of where their peers are going. It is on this model that Sanjay Joshi and his team have developed an integrated control system for robots to take into account such behavioral cues in their decision making. The research team tested the system on a small robot called marketed Evolution Robotics Scorpion. The camera of the robot was able to identify his sidekick who was ahead. Its computer system could then combine this information with behavioral indices. Rather than having programmed the robot leader to send signals directly saying we should follow, the research team sent “behavioral cues” to the second robot via a wireless device.

Teaching robots to follow men
The index said that the leader was running but did not specify in what direction. To decide the direction it should take, the robot has developed its own prediction by combining the index with other parameters such as speed and direction of the leader. The researchers concluded that robots capable of incorporating behavioral information into their decision parameters acted more efficiently by following the leader that others. “Robots increasingly capable of integrating behavioral and follow them would be easier to integrate a human work. In a hospital robot could follow such a doctor in the hallways.” A paper describing this work is published in the August 2008 issue of IEEE Transactions on Industrial Electronics.