What happened

A Waymo self-driving Jaguar I-Pace vehicle, operating in autonomous mode with a human safety driver present, struck and killed a small dog on Toland Street in San Francisco.

According to the incident report filed with the California Department of Motor Vehicles (DMV), the dog ran out into the road from behind a parked car. Waymo's initial review confirmed that the autonomous system correctly identified the dog but was "not able to avoid contact," deeming the collision "unavoidable." 

The vehicle sustained minor damage.

Why it happened

The direct cause, according to Waymo, was the sudden and high-speed movement of the off-leash dog from a blind spot (behind a parked vehicle), making the collision unavoidable despite detection. 

However, the incident highlights inherent challenges with the AI system:

• Technology: The AI system, while correctly identifying the dog, could not execute a successful evasive manoeuvre or stop in time given the dog's path and speed, suggesting an edge-case failure in prediction or actuation.

• Transparency and accountability: While Waymo filed a mandatory DMV report and issued a statement, the incident occurred amidst broader complaints from San Francisco officials about insufficient data transparency from AV companies regarding erratic driving, traffic blockages, and interference with emergency services.
  
  Waymo claimed the event was unavoidable, but without a full, publicly verifiable log of the car's sensory data, decision-making process, and the safety driver's exact input (or lack thereof), it is difficult for external parties to fully verify the company's claim.
  
  This information asymmetry places the burden of proof and ultimate safety assessment largely within the company's proprietary systems, making external accountability challenging.