A Control Pipeline for Robust Lane Keeping in Model Cars

Reliable lane keeping is a fundamental requirement of autonomous model cars. This paper presents a sophisticated image processing pipeline in combination with the Stanley steering controller to reliably steer model cars on their lane at high speeds. We show that the pipeline significantly improves robustness compared to a naive steering implementation. The results ensure reliable lane keeping and hence enable testing of higher-level algorithms for e.g. overtaking or merging scenarios.