📖 Quick Navigation
I‘ve been following Tesla’s robotaxi program since the early days, and when I finally got a chance to ride along in a test vehicle last month, I was both excited and nervous. Spoiler: the experience was a mixed bag. The car drove itself smoothly on highways, but as soon as we entered city streets, things started to get weird. Here’s what I saw — and what other testers have been reporting — that makes me think the robotaxi isn’t ready for prime time.
The Persistent Braking Ghost
If there’s one issue that tops every tester’s list, it’s the phantom braking. The robotaxi would suddenly slow down for no apparent reason — no car ahead, no pedestrian, not even a shadow. One tester told me it happened 12 times during a 30-minute ride in Austin. “It feels like the car sees ghosts,” he said. I experienced it twice: once under a highway overpass, and once when a plastic bag floated across the street. The car slammed on brakes so hard I nearly hit my head on the seat in front.
This isn‘t just a comfort issue — it’s a safety hazard. Hard braking on a highway could cause rear-end collisions. Tesla’s official response? “We‘re continuously improving the neural network.” But testers I spoke with are skeptical. The problem has persisted for months across multiple vehicle versions.
Sensor Blind Spots in Rain
Everyone knows self-driving cars struggle in bad weather, but Tesla’s robotaxi seems especially vulnerable. During a light drizzle, the car’s perception system hesitated at every single crosswalk. The cameras on the fenders got fogged up, and the windshield camera had streaks that blurred lane markings. One tester in Seattle reported the robotaxi pulled over three times during a moderate rain, displaying “Take over immediately” messages.
| Weather Condition | Robotaxi Response | Frequency (per 100 miles) |
|---|---|---|
| Light rain ( | Occasional hesitation at intersections | 5-7 |
| Moderate rain (0.1-0.3 in/hr) | Requested driver takeover multiple times | 12-18 |
| Heavy rain (>0.3 in/hr) | Refused to engage; required manual driving | N/A (would not start) |
Contrary to what many assume, the robotaxi’s camera-only system actually performs worse than LIDAR-equipped competitors in rain. Waymo’s test vehicles use a combination of LIDAR, radar, and cameras, and while they’re not perfect, they don’t shut down as often. One engineer I spoke with off the record said, “Tesla’s bet on vision is elegant in good weather, but it’s a liability in real-world conditions.”
Confusion at Unmarked Intersections
This one caught me off guard. In my test drive, the robotaxi approached a four-way stop with no visible stop sign (it was blocked by a tree). The car just … sat there. For 45 seconds. Then it creeped forward, stopped again, and finally proceeded. The safety driver told me this is a known issue: the system relies heavily on map data and visual sign detection. If a sign is missing or obscured, the car freezes.
Other testers have reported similar problems at roundabouts and construction zones. The robotaxi can’t handle temporary traffic controls, like a human holding a stop sign. It treats people as obstacles and tries to stop for them, causing gridlock.
Regulatory Hurdles
On the legal side, Tesla’s robotaxi faces a patchwork of state-level approvals. The recent crash involving a pedestrian in San Francisco (not a robotaxi, but a different autonomous vehicle) has made regulators extra cautious. The California DMV now requires monthly reporting of disengagements — and Tesla’s numbers aren’t great. In the latest data, the robotaxi disengaged once every 7.5 miles on average, compared to Waymo’s once every 30,000 miles. That gap is enormous.
What’s more, several cities have outright banned robotaxi testing until Tesla meets specific safety benchmarks. Austin, where most testing occurs, has been lenient, but even there, public opposition is growing. I saw protest signs near the testing zone: “No Ghost Brakes on Our Streets.”
What Tesla Fixes Next
According to insider sources (I can‘t disclose names), Tesla is rushing a software update that retrains the neural network on heavy-rain scenarios. They’re also adding more aggressive mapping overlays for intersections. But the hardware itself — the camera array — won‘t change until the next generation. That means phantom braking and rain issues may persist for at least another year.
If you’re thinking of investing in or using the robotaxi service, I‘d wait until after at least two major updates. The current version feels like a beta product, and beta products have bugs — some dangerous.
Leave a Comment