What Do the (Finally Public) Data on Tesla's Robotaxi Incidents Reveal?

In the Ecosystem of Smart Mobility

In the ecosystem of smart mobility, trust is built on data, not narratives. For this reason, Tesla's decision to make public all 17 previously confidential reports on accidents filed with the NHTSA (National Highway Traffic Safety Administration) represents a crucial turning point for the autonomous vehicle sector. Until recently, the Austin-based company had distinguished itself as the only entity to fully censor its accounts, systematically marking them as "confidential business information." Today, the removal of that constraint offers a new insight into the experimental Robotaxi service in Austin between July 2025 and March 2026.

A "Self-Own" of Opacity

In accordance with NHTSA General Order 2021-01, all companies engaged in the development of autonomous driving systems are required to report accidents. However, while competitors like Waymo, Zoox, Avride, and May Mobility provided detailed descriptions of each incident, Tesla's reports presented a standardized formula: [OMITTED, MAY CONTAIN CONFIDENTIAL BUSINESS INFORMATION]. The company justified this approach by citing the risk of incurring "financial damages" should competitors exploit such data to measure progress of the ADS (Automated Driving System).

This prolonged lack of transparency has resulted in a significant communication blunder. The inability to differentiate between a minor fender bender and a serious system failure fueled pointless polarizations, allowing critics and supporters to shape reality as they wished. With the latest update from NHTSA, the company has reversed its stance: the reports are now accessible and uncensored.

Accident Overview: The Constant of Human Error

The analyzed data exclusively involves 2026 Model Y vehicles equipped with ADS technology and a safety operator on board. From a statistical standpoint, the impact on physical health has fortunately been limited: the vast majority of incidents resulted in only material damage:

Type and Outcome of Accident
- Number of Accidents
  - Only material damage: 13
  - No reported injuries: 2
  - Minor injury (no hospitalization): 1
  - Minor injury (with hospitalization): 1

Analysis of the dynamics confirms a trend already widely documented by competing fleets: autonomous vehicles often fall victim to the distraction or unpreparedness of surrounding human drivers. Numerous incidents describe the Tesla being hit while it was at a complete stop - at red lights, stop signs, or aligned in traffic. Among the recorded cases are collisions with SUVs, trucks, and even the lateral impact of a city bus turning or a rickshaw hitting the side mirror. The human driver, essentially, tends to underestimate the software's strict and precise adherence to traffic laws.

Critical Issues: The Remote Human Factor and Perception Limits

While Tesla emerges diminished from accusations of its software's inherent danger in standard driving conditions, elements arise that deserve in-depth technical reflection. Two dynamics, in particular, raise concerns in the context of a driverless industrial scale-up.

The Paradox of Remote Operators:

The remote support system has shown unexpected vulnerabilities. On two separate occasions (July 2025 and January 2026), the vehicle stopped because the ADS was unable to proceed. The intervention of a remote operator, called to resolve the impasse from afar, caused the impact: in the first case, the car was driven onto a sidewalk against a metal fence at 13 km/h; in the second, it hit a construction barrier at 14 km/h. If the long-term goal is to eliminate physical safety drivers, the effectiveness of remote intervention must be radically optimized.

Perception Limitations in Obstacles:

Reports highlight persistent difficulties of the ADS in detecting geometrically complex or small obstacles, especially during low-speed maneuvers or in reverse. Collisions have been recorded against metal chains at parking lot entrances, collisions with protruding trailer hitches, and bumps against wooden light poles and curbs.

This specific type of incident reignites the debate on the strategic choice to rely solely on cameras (pure vision approach). It is reasonable to suggest that the aid of ultrasonic sensors or traditional radar could have avoided such collisions at blind spots or with thin, non-reflective obstacles.

Towards "Overly" Cautious Behavior?

The most significant incident, which required the hospitalization of the safety operator for examinations due to physical pain, occurred at only 3 km/h in a turn lane, where the Model Y was rear-ended by an SUV while yielding to cross traffic. Although formal responsibility rests entirely with the SUV driver, the incident raises a key question for future mobility: can overly cautious, hesitant, or rigidly conservative driving algorithms end up misleading human drivers, paradoxically increasing the risk of rear-end collisions?

Below Are Descriptions of the Individual Accidents:

July 2025 - Fixed object (metal fence), 13 km/h, minor injuries without hospitalization: The Tesla ADS was stationary on the right side of the road. The safety operator requested assistance because the ADS was not moving forward. The remote operator took control and drove the car left onto a sidewalk against a metal fence. No passengers were on board.
July 2025 - SUV, 0 km/h, only material damage: The Tesla ADS was stationary at a red light at an intersection. An SUV that was stopped behind the ADS slowly approached and rear-ended it. No passengers were on board the vehicle.
July 2025 - SUV, 3 km/h, minor injuries with hospitalization: The Tesla ADS was in a right turn lane with a yield sign, yielding to cross traffic. The ADS stopped, then accelerated to 3 km/h while continuing to yield and was rear-ended by an SUV that was following. The safety monitoring system driver later reported feeling pain and requested a medical evaluation.
September 2025 - Animal (dog), 43 km/h, no injuries: The Tesla ADS was driving straight in the center lane approaching a green light when a dog suddenly appeared at the intersection from the right. The ADS reduced speed and swerved left. The dog ran towards the Tesla and hit the lower part of the front right bumper, pushing it into the path of a van. The dog was later seen running away.
September 2025 - Car, 9.6 km/h, only material damage: The Tesla ADS was driving straight in a parking lot when a car reversed from a perpendicular parking lane against the signage. The ADS reduced speed and swerved left, but the reversing vehicle collided with the right side of the Tesla.
September 2025 - Fixed object (metal chain), 9.6 km/h, only material damage: The Tesla ADS made an unprotected left turn to enter a parking lot and came into contact with a metal chain. There was one passenger present.
September 2025 - Cyclist, 0 km/h, only material damage: The Tesla ADS was stationary at a red light. A rickshaw approaching from behind in the bike lane hit the right side mirror while passing it.
October 2025 - Trailer hitch, 29 km/h, only material damage: The Tesla ADS turned left onto a residential road. While passing a house, the side mirror of the ADS hit the protruding goose neck hitch of a dump trailer that extended into the road.
November 2025 - Scooter, 0 km/h, no injuries: The Tesla ADS was stationary behind traffic at a red light. A person on a scooter approached from behind, inserted themselves into the space behind the ADS, swerved right, and hit the back. The scooter driver then went over the sidewalk and continued on the sidewalk.
December 2025 - Fixed object (curb), 27 km/h, only material damage: The Tesla ADS was driving straight when the right rear tire went over a rough surface, deflating. As the ADS slowed to pull over, the flat tire hit a curb. The ADS initiated a maneuver with minimal risk. There was one passenger present.
January 2026 - Pole/tree (electric pole), 1.6 km/h, only material damage: The Tesla ADS began reversing after encountering a blocked alley. The rear hit a wooden electric pole.
January 2026 - Fixed object (curb), 3 km/h, only material damage: The Tesla ADS was reversing to enter an empty parking lot when the right rear tire hit the corner of a curb.
January 2026 - Heavy truck, 6 km/h, only material damage: The Tesla ADS was driving straight on a narrow road with a heavy tow truck parked on one side and a car parked on the other. The left side mirror of the ADS hit the platform of the tow truck. One passenger was on board.
January 2026 - Bus, 0 km/h, only material damage: The Tesla ADS was stationary at a red light in the leftmost lane (the adjacent right lane was blocked by construction barriers). An urban bus turning right from the perpendicular street collided laterally with the ADS with its front bike rack. Two passengers were on board.
January 2026 - Fixed object (construction barrier), 14 km/h, only material damage: The safety monitor requested navigation assistance. A remote operator took control when the ADS stopped and violently hit a temporary construction barrier, scraping the left front fender and tire.
March 2026 - Car, 0 km/h, only material damage: The Tesla ADS was stopped at an intersection in the left turn lane with a flashing yellow arrow. A car stopped behind the ADS started moving and rear-ended it.
March 2026 - Pickup, 0 km/h, only material damage: The Tesla ADS was stopped at a stop sign. A truck following the ADS started moving and rear-ended it. A passenger was on board.