Table 2.

System comparison between our work and Stanford's supervised model.

Aspect	Details	Our agent	Stanford's agent
Skills	Turning left/right, driving forward	✓	✓
	Driving backward	✓
	Brake	✓
	Turning around	✓
Bad situation	Driving reverse		✓
	Getting stuck		✓
Performance comparison	Less time-consuming		✓
	Finishing the tracks	✓	✓
Data-consuming	Handcrafted dataset		✓
	Annotation		✓
	Unlabeled data	✓
Self-adaptability		✓
Online learning paradigm		✓