My Experience with Waymo One

Waymo One is the ride-hailing service launched by Waymo. To my knowledge, it is the only commercial autonomous taxi service that operates without a safety driver behind the wheel. The company started testing its autonomous Chrysler Pacificas without a safety driver, in mid 2017. In 2018, the company started offering autonomous rides as a commercial service, named Waymo One.

I’ve always wanted to ride in one of these, but laziness kept me confined to my house. Finally, when my brother came to visit, I thought I’d get out of my abode, and show him what the future holds with regard to mobility.

We parked our car at the nearest parking lot we could find and planned a round trip so that the Waymo Driver could drop us back to our car. Our start and end points are shown in the following map.

The trip from Anytime Fitness to Walmart. The round trip is around 20.4 miles and 31 minutes long.

As one can see, it’s an approximately 20 mile round trip, and Google Maps shows the duration to be around 31 minutes at the time of writing this post. Here’s a screenshot from my Waymo One trip history.

As shown above, the distance was slightly longer, and the trip completion time was much greater. This is possibly because the Waymo Driver avoided the freeway (AZ-101 Loop), and the fact that it acted too conservatively.

Overall, I was very impressed with the experience. At first, the car arrives to pick you up, with your initials on a slick LED display. Once you press the start button, your ride starts, and the steering wheel starts moving on its own. Check the following video.

The vehicle is also able to safely navigate through cluttered environments, with a lot of live pedestrians, without significantly disturbing the pedestrians. We can see the detected pedestrians on the GUI in the car. Check this next video to find out.

The cost of the round trip was $30.23. UberX shows the cost to be $14.32 for a one-way trip, Lyft showing $25.94 (this might be surge pricing). Cost-wise, Waymo One seems to be competitive with existing commercial ride-hailing services. While the ride was relatively smooth and safe, here are some observations I’d like to share:

1. The Waymo Driver was very conservative in its route planning — As shown above, the Waymo Driver (henceforth referred to as WD) took us along a slighly longer distance. This is probably because the WD chose to travel through residential areas, to avoid traffic, freeways, and difficult left turns. Also, it didn’t move when there were pedestrians standing close to it. In those cases, a human driver would have slowly moved to the destination, but the WD didn’t.
2. The vehicle came to an abrupt stop once on the street, and a few other times in the parking lot. It was obviously adhering to strict safety standards, but a human driver would’ve provided a much smoother experience.
3. Cost-wise, it’s comparable to Uber and Lyft. Uber cost $14 for a one-way trip, while the WD cost around $30 for a round trip.
4. The speed of the car was very precisely controlled, and it was able to handle variations in road traffic quite smoothly. However, there was this one UPS van who had its emergency lights flashing, and the WD drove closer to it than I would’ve.
5. The visualization tool was great, it provided both car and map views. In the car view, you could visualize some of the perception outputs — object detections and pointclouds, and they seemed to match closely with the ground truth. You can also see the planned trajectory, and that the WD follows the planned trajectory with remarkable precision
6. In one incident, the WD wanted to change to the left lane, but the lady in the left lane wouldn’t let it do it (presumably to challenge the vehicle) until a signal. After the signal turned green, the AV accelerated and changed the lane before the lady could obstruct it again.
7. At the side of the car, there was a spinning Lidar sensor that didn’t stop even though I brought my finger really close to it. There’s a chance that a child could bring their finger close to it and get injured. (I know I’m nitpicking here, but I’ll put it out there anyway.)

I want to reiterate that despite the above points, this solution is still very impressive, and it works, despite having to deal with silly, unpredictable humans! The fact that Waymo is confident enough to deploy this as a commercial product speaks volumes about its safety. However, it still has a long way to go before it becomes a commercially viable product, due to the reasons listed above. It’s a challenging task, but I’m sure Waymo is driven to accomplish it.