Scaling Planning for Automated Driving using Simplistic Synthetic Data

We challenge the perceived consensus that the application of deep learning to solve the automated driving planning task requires a huge amount of real-world data or a realistic simulator. Using a roundabout scenario, we show that this requirement can be relaxed in favour of targeted, simplistic simulated data. A benefit is that such data can be easily generated for critical scenarios that are typically underrepresented in realistic datasets. By applying vanilla behavioural cloning almost exclusively to lightweight simulated data, we achieve reliable and comfortable real-world driving. Our key insight lies in an incremental development approach that includes regular in-vehicle testing to identify sim-to-real gaps, targeted data augmentation, and training scenario variations. In addition to the methodology, we offer practical guidelines for deploying such a policy within a real-world vehicle, along with insights of the resulting qualitative driving behaviour. This approach serves as a blueprint for many automated driving use cases, providing valuable insights for future research and helping develop efficient and effective solutions.