Robotaxis: Waymo and Tesla make real progress, glitches included

Waymo has put real driverless cars on the road for over two years now. It's not a controlled test in a parking lot, these are actual vehicles picking up passengers in San Francisco and Phoenix every day. Waymo logged over 2 million autonomous miles in 2024, a number that sounds impressive until you remember they're concentrated on tested routes with stable weather.

Tesla claims it has the most advanced tech, but the Cybercab remains a promise without concrete deliveries. The gap between what they show in videos and what works on a rainy street corner is still massive. Baidu pushes hard in China with hundreds of vehicles, but regulatory context there differs vastly from Europe and North America.

Current robotaxis suffer from an embarrassing weakness: heavy rain and snow confuse the sensors. A human driver turns on wipers and moves on, an autonomous car slows down and waits. In cities like Boston or Seattle that's a serious problem, not a minor quirk.

According to leaked internal reports, Waymo slows significantly with chaotic traffic and unpredictable human behavior (and we know how people drive). Edge cases still number in the dozens, from unmarked lanes to broken traffic lights to drunk people crossing randomly. Each scenario requires extra training hours for the system.

There's also infrastructure behind it: ultra-precise maps, remote control centers for critical situations, constant maintenance. It's not just smart software, it's an entire ecosystem of human backup support that keeps passenger prices high.

The realistic timeline isn't tomorrow. Waymo covers optimal conditions today: well-mapped streets, known urban zones, predictable weather. Expanding to secondary cities or extreme climates requires years of extra work, not months.

Tesla promises everything now, but history teaches caution. The company downscaled expectations before on autopilot and full self-driving, it could happen again. Meanwhile regulations evolve slowly and insurers still don't know how to price a car making autonomous decisions.

For the average person, robotaxis will arrive in major cities in controlled forms within 3-5 years, not as total replacement. They'll be an extra option for certain routes, not the immediate revolution headlines promise. The trade-off is already written: they work well in specific conditions, period.