<ul><li>End-to-end vision-based imitation learning in autonomous driving has shown promise, but traditional approaches lack confidence estimation and precision.</li><li>A dual-head neural network architecture is introduced that combines regression and classification heads to improve decision reliability.</li><li>The regression head predicts continuous driving actions, while the classification head estimates confidence for corrections in low-confidence scenarios.</li><li>Experimental results demonstrate improved driving stability, reduced lane deviation, and enhanced trajectory accuracy compared to regression-only models.</li></ul>

CARIL: Confidence-Aware Regression in Imitation Learning for Autonomous Driving

Discover more