Brandon Basso describes the challenge of making an AI driver capable across varied conditions and demonstrating that it behaves safely. He separates safety, comfort, rule compliance and competence as related evaluation concernsEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria..
He discusses geographic diversity in driving data, including temporary roadworks and local conventions, and describes using collected, licensed and simulated data.
Basso explains Wayve’s approach to learning safety-related representations within its driving system and discusses a possible future transition from supervised assistance to eyes-off operation. He presents timing and benefits as expectations, not established guarantees.
Watch on YouTube
