Vision-based avoidance of obstacles with unknown constant velocity

This paper discusses vision-based avoidance of obstacles with unknown sizes and unknown constant velocities for four-wheel nonholonomic robots. The robot is equipped with an inertial measurement unit that provides measurements of the robot´s position and velocity, and a monocular camera that detects the obstacles. An analytical scheme is adopted for trajectory generation and vehicle control. Vision sensor is used to obtain the estimates of the obstacles´ unknown sizes and unknown velocities that are used by the trajectory generation scheme. The effectiveness of the proposed technique is demonstrated by numerical simulation examples.