Simple, real-time obstacle avoidance algorithm for mobile robots

This paper proposes a novel, reactive algorithm for real time obstacle avoidance, compatible with low cost sonar or infrared sensors, fast enough to be implemented on embedded microcontrollers. We called this algorithm "the bubble rebound algorithm". According to this algorithm, only the obstacles detected within an area called "sensitivity bubble" around the robot are considered. The shape and size of the sensitivity bubble are dynamically adjusted, depending on the kinematics of the robot. Upon detection of an obstacle, the robot "rebounds" in a direction having the lowest density of obstacles, and continues its motion in this direction until the goal becomes visible, or a new obstacle is encountered. The performances and drawbacks of the method are described, based on the experimental results with simulators and real robots.