collisionavoidance

Collision avoidance refers to methods and systems that detect potential collisions and act to prevent them. It spans transportation, robotics, and automation, and covers both proactive planning and reactive control. The aim is to minimize risk while maintaining task performance.

Sensing uses radar, LiDAR, cameras, sonar, and ultrasonic sensors, sometimes with GPS and vehicle-to-everything communications. Prediction estimates other objects’ trajectories. Planning determines a safe path or velocity, and control enacts the maneuver within constraints and safety margins.

Techniques range from classical to hybrid. Global planning uses maps and objectives to produce a collision-free route; local and reactive methods adjust to dynamics. Notable approaches include potential fields, velocity obstacles and reciprocal velocity obstacles, dynamic window, and model predictive control. Data-driven and probabilistic methods address sensing uncertainty.

Applications cover autonomous road vehicles, aerial drones, ships, industrial robots, and human-robot interaction. In aviation and maritime contexts, TCAS and COLREGs guide avoidance. In networks, collision-avoidance concepts underlie access protocols like CSMA/CA, which minimize simultaneous transmissions.

Key challenges include uncertain sensing, latency, multi-agent coordination, safety guarantees, verification, and regulation. Real-time performance, human factors, and balancing safety with efficiency remain active research areas, with ongoing work on standards and robust perception under adverse conditions.

See also obstacle avoidance, autonomous navigation, collision avoidance system, TCAS, COLREGs, and CSMA/CA.