Understanding the intent of a cyclist doing a track-stand is a wrinkle for self-driving cars to figure out. (Eric Risberg/AP)

Earlier this month in Austin, a cyclist and a Google self-driving car met at a four-way stop. This likely wasn’t the first time a Google self-driving vehicle has encountered a cyclist at a four-way stop. The company’s vehicles have driven more than 1.1 million miles in autonomous mode.

But the encounter featured a twist — the cyclist was doing a track stand.

The cyclist recounted the encounter on an online bike forum:

The car got to the stop line a fraction of a second before I did, so it had the [right of way]. I did a track-stand and waited for it to continue on through.

But the track stand, which are generally done only by riders on fixed-gear bikes, quickly became a problem.

[Look inside Google’s new self-driving car]

The self-driving cars are notoriously careful, and tend to brake when anyone else is moving forward into the vehicle’s path. In a track stand, a rider on a fixed-gear bike may shift ever so slightly forward and back in an effort to maintain balance. (Watch video of a track stand here.) Also, a rider doing a track stand maintains the body position typical of a cyclist in motion, not one that is stopping. For riders of fixed-gear bikes, it can be a fun game to never have to put one’s foot down on the pavement, but instead balance at stop signs and red lights.

While a human driver can easily see a rider doing a track stand isn’t going anywhere, Google’s self-driving car seems to be still be figuring that out.

As the cyclist recalled:

It apparently detected my presence … and stayed stationary for several seconds. it finally began to proceed, but as it did, I rolled forward an inch while still standing. The car immediately stopped…
I continued to stand, it continued to stay stopped. Then as it began to move again, I had to rock the bike to maintain balance. It stopped abruptly.
We repeated this little dance for about two full minutes and the car never made it past the middle of the intersection. The two guys inside were laughing and punching stuff into a laptop.
Despite the awkward encounter, the cyclist didn’t leave with a negative impression of self-driving cars.

“The odd thing is,” wrote the cyclist, “I felt safer dealing with a self-driving car than a human-operated one.”

Self-driving cars could be a boon for cycling. If self-driving vehicles almost never crash, roads will become immensely more safe and inviting to cyclists. But for now, mastering how to interact with cyclists is a challenge for self-driving vehicles.

A patent Google received this spring detailed how its self-driving cars could identify cyclists and interpret their hand signals. It also mentioned the ability to identify a cyclist by measuring the distance between the pavement and the top of a stopped cyclist’s head.

Of course, until this summer Google’s testing was centered near its Mountain View, Calif., headquarters. In July, Google began testing in Austin, home to a lot more hipsters and fixed-gear bikes.

A Google spokeswoman called the situation a good example of the feedback Google wants to get from the community as its refines its software and tests in areas outside Mountain View.

The broader the experiences of self-driving vehicles, the better prepared they will be for real-world driving. The run-in also highlights the long list of rare situations the cars will have to master before they can replace human drivers. After all, what happens when a self-driving car approaches a downtown intersection with multiple cyclists on fixed-gear bikes, and a herd of pedestrians?

In a July report on emerging technologies, the research firm Gartner placed autonomous vehicles at the peak of inflated expectations. It wrote that inflated and unrealistic expectations have resulted from prominent media reports, despite complexity and cost challenges that the vehicles face.

Related: The race is on to figure out what self-driving cars should look like