LiDAR-as-Camera for End-to-End Driving
This addresses sensor alignment issues for autonomous driving systems, offering a practical solution with improved robustness, though it is incremental as it builds on existing depth-sensing benefits.
The paper tackled the challenge of sensor alignment in end-to-end autonomous driving by using LiDAR-images from Ouster LiDARs, which provide perfectly aligned depth and visual data, and demonstrated that these perform at least equally to camera-based models in real-car road-following, with better generalization to new weather conditions.
The core task of any autonomous driving system is to transform sensory inputs into driving commands. In end-to-end driving, this is achieved via a neural network, with one or multiple cameras as the most commonly used input and low-level driving command, e.g. steering angle, as output. However, depth-sensing has been shown in simulation to make the end-to-end driving task easier. On a real car, combining depth and visual information can be challenging, due to the difficulty of obtaining good spatial and temporal alignment of the sensors. To alleviate alignment problems, Ouster LiDARs can output surround-view LiDAR-images with depth, intensity, and ambient radiation channels. These measurements originate from the same sensor, rendering them perfectly aligned in time and space. We demonstrate that such LiDAR-images are sufficient for the real-car road-following task and perform at least equally to camera-based models in the tested conditions, with the difference increasing when needing to generalize to new weather conditions. In the second direction of study, we reveal that the temporal smoothness of off-policy prediction sequences correlates equally well with actual on-policy driving ability as the commonly used mean absolute error.