Enhanced Radar Perception via Multi-Task Learning: Towards Refined Data for Sensor Fusion Applications
This work addresses a specific bottleneck in radar perception for autonomous vehicles and sensor fusion applications, representing an incremental improvement over existing height extension methods.
This paper tackles the problem of radar point clouds lacking height information by introducing a learning-based approach to infer height, reducing average absolute height error from 1.69 to 0.25 meters compared to state-of-the-art methods. The refined radar data enhances downstream perception tasks like object detection and depth estimation in radar-camera fusion models.
Radar and camera fusion yields robustness in perception tasks by leveraging the strength of both sensors. The typical extracted radar point cloud is 2D without height information due to insufficient antennas along the elevation axis, which challenges the network performance. This work introduces a learning-based approach to infer the height of radar points associated with 3D objects. A novel robust regression loss is introduced to address the sparse target challenge. In addition, a multi-task training strategy is employed, emphasizing important features. The average radar absolute height error decreases from 1.69 to 0.25 meters compared to the state-of-the-art height extension method. The estimated target height values are used to preprocess and enrich radar data for downstream perception tasks. Integrating this refined radar information further enhances the performance of existing radar camera fusion models for object detection and depth estimation tasks.