论文标题
支柱：自动驾驶的端到端鸟眼视图估算
PillarFlow: End-to-end Birds-eye-view Flow Estimation for Autonomous Driving
论文作者
论文摘要
在自动驾驶中，准确估算周围障碍的状态对于安全和健壮的路径计划至关重要。但是，由于外观和遮挡变化，这种感知任务很困难，尤其是对于通用障碍物/对象而言。为了解决这个问题，我们为Bird's Eye View（BEV）中的基于LiDAR的流量估算提出了一个端到端的深度学习框架。我们的方法将连续的点云对作为输入，并产生一个2-D BEV流网格，描述了每个单元的动态状态。实验结果表明，所提出的方法不仅可以准确估计2-D BEV流，还可以改善动态和静态对象的跟踪性能。
In autonomous driving, accurately estimating the state of surrounding obstacles is critical for safe and robust path planning. However, this perception task is difficult, particularly for generic obstacles/objects, due to appearance and occlusion changes. To tackle this problem, we propose an end-to-end deep learning framework for LIDAR-based flow estimation in bird's eye view (BeV). Our method takes consecutive point cloud pairs as input and produces a 2-D BeV flow grid describing the dynamic state of each cell. The experimental results show that the proposed method not only estimates 2-D BeV flow accurately but also improves tracking performance of both dynamic and static objects.
