Part 6/14:
He contrasted this with traditional modeling, which is error-prone and hard to tune, especially when codifying moral or ethical trolley problems, like avoiding wildlife or choosing whether to go into oncoming traffic to bypass an obstacle.
Handling Difficult Corner Cases and Edge Scenarios
Tesla’s vast data pipeline enables learning from rare but critical corner cases, which are essential for robust autonomy. Examples include:
Chickens crossing the road: A humorous but illustrative case showcasing the vehicle's patience and understanding of intent.
Stationary geese in a parking lot: The system recognizes non-moving animals and proactively maneuvers around them.