Abstract: Knowledge distillation in autonomous driving typically involves transferring information from high-cost sensor modalities such as LiDAR to camera-based models to improve 3D object detection.