Deploying autonomous robots in crowded indoor environments usually requires them to have accurate dynamic obstacle perception. Although plenty of previous works in the autonomous driving field have investigated the 3D object detection problem, the usage of dense point clouds from a heavy LiDAR and their high computation cost for learning-based data processing make those methods not applicable to lightweight robots, such as vision-based UAVs with small onboard computers. To address this issue, we propose a lightweight 3D dynamic obstacle detection and tracking (DODT) method based on an RGB-D camera. Our method adopts a novel ensemble detection strategy, combining multiple computationally efficient but low-accuracy detectors to achieve real-time high-accuracy obstacle detection. Besides, we introduce a new feature-based data association method to prevent mismatches and use the Kalman filter with the constant acceleration model to track detected obstacles. In addition, our system includes an optional and auxiliary learning-based module to enhance the obstacle detection range and dynamic obstacle identification. The users can determine whether or not to run this module based on the available computation resources. The proposed method is implemented in a lightweight quadcopter, and the experiments prove that the algorithm can make the robot detect dynamic obstacles and navigate dynamic environments safely.

翻译：在拥挤的室内环境中部署自主机器人通常需要具备精确的动态障碍物感知能力。尽管自动驾驶领域已有大量工作研究了三维目标检测问题，但重载激光雷达生成的高密度点云以及基于学习的数据处理方法带来的高计算成本，使得这些方法无法适用于轻量级机器人（例如配备小型机载计算机的视觉无人机）。为解决这一问题，我们提出了一种基于RGB-D相机的轻量级三维动态障碍物检测与跟踪（DODT）方法。该方法采用新颖的集成检测策略，通过组合多个计算高效但精度较低的检测器，以实现实时高精度障碍物检测。此外，我们引入了一种基于特征的新型数据关联方法以防止误匹配，并采用恒定加速度模型的卡尔曼滤波器对检测到的障碍物进行跟踪。同时，系统包含一个可选且辅助性的学习模块，用于增强障碍物检测范围与动态障碍物识别能力。用户可根据可用计算资源决定是否运行该模块。所提方法已在轻量级四旋翼飞行器上实现，实验证明该算法能使机器人检测动态障碍物并在动态环境中安全导航。