obstacleavoidance

Obstacle avoidance refers to techniques that enable an agent to detect obstacles in its path and maneuver to avoid collisions while pursuing a goal. It spans perception, planning, and control subsystems used in robots, autonomous vehicles, drones, and other mobile systems. The objective is to achieve safe, efficient travel or manipulation in environments that may be static or dynamic and may change over time.

Perception gathers data from sensors such as cameras, LiDAR, radar, sonar, and depth sensors, then processes and fuses information to identify obstacles, estimate their position, size, and velocity, and build a local map. Robust obstacle detection must cope with noise, occlusion, and changing lighting or weather, often using probabilistic or machine learning methods.

Planning and control separate global and local tasks. Global planners compute a route that avoids known obstacles, while local or reactive planners adjust the trajectory in real time to avoid unforeseen obstacles using methods such as potential fields, velocity obstacles, dynamic window, model predictive control, or sampling-based planning like A*, RRT*. Receding horizon approaches are common for dynamic environments.

Applications include autonomous cars, delivery robots, warehouse automation, aerial drones, service robots, and robotic arms. Challenges include real-time computation, sensor fusion reliability, dynamic and crowded environments, multi-agent coordination, and ensuring safe behavior under uncertainty. Evaluation often uses simulation and real-world benchmarks to compare safety, efficiency, and robustness.