Has Autonomous Driving Crashed Again? Test Vehicle in Austin Kills a Viral Duck

After taking 3 minutes to get the full picture, I’m still a bit fired up. Recently, near Mueller Lake Park in Austin, Texas, a self-driving car from Avride that was undergoing testing was accused of running over and killing a locally beloved “internet-famous duck,” and the incident quickly inflamed community sentiment. For many people, this was not just a simple accident involving a small animal, but more like the most basic command in a turn-based game that absolutely should not have been missed—yet the system somehow still produced the worst possible answer.

According to a witness in a local Facebook group, the Avride vehicle drove past the park earlier this week and ran directly over a duck that was resting there. Even more unsettling, the witness said there was someone in the driver’s seat, but their hands were not on the steering wheel. The vehicle neither slowed down nor showed any obvious hesitation when it struck the duck, and afterward, the people inside did not stop to check what had happened, but simply drove away. According to the witness, the duck also appeared to be lying next to a nest of eggs at the time, making the whole scene especially upsetting to imagine.

In addition, the witness said that before hitting the duck, the car had run a stop sign and nearly collided with his vehicle. If that detail is ultimately confirmed by the investigation, then the issue is no longer just about insufficient ability to identify small animals, but about multiple warning lights flashing at once across testing procedures, human takeover, and road safety. Unfortunately, although autonomous driving technology has continued to improve over the years in recognizing pedestrians, vehicles, and road conditions, the industry as a whole still has clear shortcomings when it comes to smaller, more complex targets such as cats, dogs, chickens, and ducks. And Avride is not alone—some other brands have previously also been involved in incidents where small animals were struck by mistake.

As public attention intensified, Avride’s response was fairly standard: the company said it takes safety very seriously and is conducting a full internal investigation into the incident. The company also stated that the vehicle’s design complies with relevant traffic regulations and is capable of recognizing and responding to road conditions, road users, and obstacles through its onboard systems, while also researching dedicated technical solutions to avoid similar incidents as much as possible in the future. Notably, Avride has temporarily suspended testing operations on certain roads around Mueller Lake, which at least shows the company understands that the most important thing right now is not to stubbornly argue its case, but to get the risks under control first.

Judging from community feedback, many local residents have already begun calling for restrictions on autonomous vehicles and robotaxis operating on public roads. That reaction is actually easy to understand: tech companies talk about system capabilities, but what residents feel is the real risk right outside their homes. The gap between the two is like the difference in a JRPG between impressive-looking stats on paper and actual combat performance—no matter how good the parameters look, they cannot withstand one jarring crash. For the industry, this incident will most likely once again bring an old question back to the forefront: how exactly should autonomous driving handle biological recognition scenarios involving “small targets, low postures, and irregular movement”?

Still, this incident also serves as a reminder to onlookers that discussions about autonomous driving cannot focus only on whether the car can run, but also on whether it can remain reliable in edge-case scenarios. Recognizing large obstacles is only the basic level; what truly determines public perception is often these seemingly minor moments that are actually closest to everyday life. For Avride, the outcome of the investigation, the clarification of responsibility, and the subsequent technical improvements will be the key to whether this incident can truly be put behind it.