This paper introduces the dynamic window approach for reactive collision avoidance in mobile robots equipped with synchro-drives. The approach is derived directly from the robot's motion dynamics, making it particularly suitable for high-speed robots. Unlike previous methods, it searches for commands in the velocity space, reducing the search space to the dynamic window, which consists of velocities reachable within a short time interval. Within this window, only admissible velocities that allow the robot to stop safely are considered. The optimal velocity combination is chosen by maximizing an objective function that includes progress towards the goal, forward velocity, and distance to the next obstacle. Extensive experiments with the robot RHINO, operating at speeds up to 95 cm/sec, demonstrate the approach's effectiveness in various environments, including cluttered and dynamic settings. The method has been successfully applied using ultrasonic sensors and, in some cases, cameras and infrared detectors. The dynamic window approach ensures robust and elegant collision avoidance, making it suitable for low-cost robots with limited motor torques.This paper introduces the dynamic window approach for reactive collision avoidance in mobile robots equipped with synchro-drives. The approach is derived directly from the robot's motion dynamics, making it particularly suitable for high-speed robots. Unlike previous methods, it searches for commands in the velocity space, reducing the search space to the dynamic window, which consists of velocities reachable within a short time interval. Within this window, only admissible velocities that allow the robot to stop safely are considered. The optimal velocity combination is chosen by maximizing an objective function that includes progress towards the goal, forward velocity, and distance to the next obstacle. Extensive experiments with the robot RHINO, operating at speeds up to 95 cm/sec, demonstrate the approach's effectiveness in various environments, including cluttered and dynamic settings. The method has been successfully applied using ultrasonic sensors and, in some cases, cameras and infrared detectors. The dynamic window approach ensures robust and elegant collision avoidance, making it suitable for low-cost robots with limited motor torques.