Visuomotor Understanding for Representation Learning of Driving Scenes
This work addresses the need for detailed scene understanding in autonomous driving by leveraging naturally paired data, though it is incremental as it builds on existing self-supervised learning approaches.
The paper tackles the problem of learning visual representations from unlabeled driving videos paired with vehicle sensor data, using a self-supervised framework to predict dense optical flow from single frames, and shows that this representation outperforms unsupervised methods on semantic segmentation.
Dashboard cameras capture a tremendous amount of driving scene video each day. These videos are purposefully coupled with vehicle sensing data, such as from the speedometer and inertial sensors, providing an additional sensing modality for free. In this work, we leverage the large-scale unlabeled yet naturally paired data for visual representation learning in the driving scenario. A representation is learned in an end-to-end self-supervised framework for predicting dense optical flow from a single frame with paired sensing data. We postulate that success on this task requires the network to learn semantic and geometric knowledge in the ego-centric view. For example, forecasting a future view to be seen from a moving vehicle requires an understanding of scene depth, scale, and movement of objects. We demonstrate that our learned representation can benefit other tasks that require detailed scene understanding and outperforms competing unsupervised representations on semantic segmentation.