A faster collision detection algorithm can allow robots to work more quickly and efficiently, according to engineers from the University of California at San Diego. The algorithm, called Fastron, works 8 times faster than existing collision detection algorithms. He uses machine learning to help robots move in complex and rapidly changing environments in real time.

Experts assume that the Fastron algorithm will be useful for robots working with people, since in this environment, machines are dealing with moving objects and must move smoothly. Fastron can also be used for robots used in homes and in the games and cinema industry. Currently, researchers are studying how the new algorithm can improve the operation of the da Vinci surgical system, in which the robotic arm autonomously performs auxiliary tasks without interfering with surgeons.

The Fastron algorithm creates a model of the configuration space of the robot, or C-space, which is a space that shows all possible positions that a robot can occupy. Fastron simulates a C-space using only a sparse set of points, consisting of a small number of so-called collision points and points without collisions. Then the algorithm determines the boundary between the points of collision and the points without collision, creating a diagram depicting abstract obstacles in C-space.