Скачать или смотреть mmWave RADAR and CBF power obstacle avoidance for drones

We demonstrate robust radar-based obstacle avoidance for drones, leveraging radar’s lightweight design, thin-obstacle detection, and full-day/night operability compared to stereo cameras and 3D LiDAR. Obstacle avoidance is formulated as a quadratic program using a Control Lyapunov Function–Control Barrier Function (CLF-CBF) framework. Within this, our Depth-Control Barrier Function (Depth-CBF) computes the buffered minimum squared distance to obstacles in the radar point cloud, with gradients estimated via a local quadratic fit and efficiently queried through a k-d tree. This approach enables real-time, all-weather collision avoidance applicable to a wide range of aerial robotics platforms. ​