Improved Optimal Reciprocal Collision Avoidance Algorithm in Racing Games

Abstract

Autonomous character behavior design is a key to the development of game artificial intelligence. Currently, collision avoidance technology, which is widely used in the field of robotics, cannot be directly applied to the autonomous character behavior design in racing games without specific adjustment. This paper makes some improvements to the classical Optimal Reciprocal Collision Avoidance (ORCA) algorithm to propose a novel collision avoidance algorithm (IORCA) suitable for racing games. Relevant concepts unique to racing games are put forward. The collision handling principles followed by racing AI on the straight lines and curves are further presented. By implementing the two systems equipped with ORCA and IORCA, the collision avoidance behavior of the racing AI in multiple scenarios are compared. It could be found that the collision avoidance behavior under IORCA is more suitable for racing games, and is able to give players a better gaming experience.