brain/
← all entities
entitygenericautonomous-driving

Tesla FSD

Notes

Tesla FSD

One-line summary: Tesla's vision-only driver-assist and robotaxi stack — supervised Level 2 for consumers, company-labeled "unsupervised" (remote-supervised) robotaxi now listed in seven US metros as of the Q2 2026 Update.

What it is

Tesla's "Full Self-Driving (Supervised)" is the company's end-to-end neural-network driving system running on either HW3 or AI4 (formerly HW4) in-car compute. v13 introduced a transformer-based "world model" with a ~10-second temporal buffer and voxelized 3D scene representation. v14 (rolling out 2026) pushes further on AI4, with a "v14 Lite" planned for HW3 2026-04-20-autoresearch-tesla-fsd.

The same stack powers Tesla's Robotaxi service. As of the April 2026 source it was unsupervised (no human in front seat, remote monitoring only) in Austin, Dallas, and Houston 2026-04-20-autoresearch-tesla-fsd. The Q2 2026 Update (posted 22 July) lists seven US metros: Texas and Florida "Ramping Unsupervised," Phoenix and Las Vegas "Preparations Underway," and the Bay Area still FSD (Supervised) under CA TCP # TCP0046782 – A 2026-08-24-autoresearch-tesla-fsd-robotaxi-jul-aug-2026.

Why it matters to autonomous-driving

Tesla is the most visible counter-bet to the Waymo/sensor-fusion school: it stakes a full consumer-and-robotaxi business on vision-only perception at scale. Whether that bet works or fails is one of the defining experiments of the field.

Key facts

  • Hardware generations: HW3 (legacy), AI4 (current), AI5 (delayed to early 2027, ~9× HW4 memory per Musk, January 2026) 2026-04-20-autoresearch-tesla-fsd.
  • Sensor stack: Cameras only. No LiDAR, no radar. Unique among major OEMs 2026-04-20-autoresearch-tesla-fsd.
  • SAE level: Level 2 (supervised). California's CPUC explicitly states Tesla "is not operating an autonomous vehicle service" 2026-04-20-autoresearch-tesla-fsd.
  • Unsupervised fleet (April 2026): Austin ~4–8 Model Ys of ~42 total; Dallas and Houston ~25 sq mi geofences each, launched April 18, 2026 2026-04-20-autoresearch-tesla-fsd.
  • Robotaxi coverage (Q2 2026 Update, 22 July): seven US metros. Austin / Dallas / Houston / Miami / Orlando / Tampa = "Ramping Unsupervised"; Phoenix / Las Vegas = "Preparations Underway"; SF Bay Area = safety-driver / FSD (Supervised) under CA TCP # TCP0046782 – A. Cybercab started production; public-road engineering test drives in the quarter; employee campus rides at Gigafactory Texas in July. Active FSD subscriptions 1.48M (+56% YoY). v14 Lite began delivering on AI3 in the US and South Korea (July) — a distillation of v14, not a public v15 consumer rollout 2026-08-24-autoresearch-tesla-fsd-robotaxi-jul-aug-2026.
  • Company Q2 call claims (22 July; first-party, unaudited). Ashok Elluswamy: 380,000 unsupervised miles across six cities in two states; "zero notable incidents" / "impeccable safety record"; robotaxi fleet already on early FSD v15 (~40% of seven planned tracks merged); cameras-only "validation" vs LiDAR/radar/HD maps. Musk: Cybercab needs chassis-specific miles before volume; scaling constraint is the "March of 9s"; "if we injure even one person… regulators will immediately clamp down." Treat as company claims — see fsd-safety-data-validity 2026-08-24-autoresearch-tesla-fsd-robotaxi-jul-aug-2026.
  • Nevada AVNC (20 August 2026). NTA unanimously approved Tesla Robotaxi, LLC to operate as an Autonomous Vehicle Network Company: paid rides in "fully autonomous" vehicles in Clark County, up to 5,000 vehicles in the first 12 months, 120-day commence-ops clock, inspections/insurance/rates required before public service. Airport service needs separate Clark County aviation authorization. Official source is the Nevada DBI press release, not the nta.nv.gov order PDF (that PDF was not retrieved). Tesla's 22 July Update still listed Las Vegas as preparations-only 2026-08-24-autoresearch-tesla-fsd-robotaxi-jul-aug-2026.
  • China DMS recall (21 August 2026) — ADAS supervision, not hardware egress. SAMR S2026M0039I: 2,740,642 domestic Model 3/Y (4 Mar 2019–7 Dec 2025). Defect: existing attention monitoring insufficient to prompt when gaze leaves the road under combination driving assistance. Remedy: OTA adding cabin-camera monitoring on top of steering-torque. The same-day door-handle recall (2,975,910) is auto hardware / occupant-egress and is not an autonomy fact — it lives on the stock-market tesla page 2026-08-24-autoresearch-tesla-fsd-robotaxi-jul-aug-2026.
  • Community-tracker disengagement rate: city miles to critical disengagement fell from ~4,109 (v14.1 peak, October 2025) to ~809 after v14.2 2026-04-20-autoresearch-tesla-fsd. No fetched v15 community-tracker update in the August pass.
  • HW3/AI4 gap (community data): AI4 ~450 miles between critical disengagements vs. HW3 ~120 miles 2026-04-20-autoresearch-tesla-fsd.
  • Oct 1, 2026 next-gen Roadster (not an FSD/robotaxi datapoint). Tesla locked a Thursday October 1, 2026 Roadster reveal (Tesla/Musk X 10.01). Stock-market tesla files it as a halo / brand event. It does not update unsupervised miles, paid robotaxi rides, FSD attach, NHTSA EA26002, or the 1.48M subscription count. Treat as adjacent Tesla calendar, not a stack/safety/ops print. From 2026-09-21-tsla-next-gen-roadster-reveal-oct-1-2026-pre-event-brief.

Strengths (from our perspective)

  • Operates on arbitrary streets rather than an HD-mapped ODD — wider domain than mercedes-drive-pilot.
  • Fleet-learning flywheel: large installed base of camera-equipped vehicles feeding training data.
  • First-person operator assessment from Karpathy (post-tenure, October 2025). andrej-karpathy led Tesla AI / Autopilot 2017–2022. He still rates Tesla's strategy as the more scalable bet relative to lidar-heavy alternatives, with the caveat that he's not fully independent. andrej-karpathy in 2025-10-17-dwarkesh-patel-andrej-karpathy-summoning-ghosts: "I just think people again are sometimes a little bit too naive about some of the progress and I still think there's a huge amount of work and I think Tesla took in my mind a lot more scalable approach and I think the team is doing extremely well... But I do think Tesla is taking the more scalable strategy and it's going to look a lot more like that." Hold as one weighted operator view, not consensus.

Weaknesses (from our perspective)

Open questions

Sources

Related

Referenced by
Concepts
Demo-to-product gap: the "march of nines"FSD Safety Data ValidityVision-Only vs Sensor Fusion
Questions
Does remote-supervised robotaxi qualify as driverless?Is vision-only perception sufficient for L4 autonomy?Will NHTSA EA26002 trigger an FSD recall?
Entities
Andrej KarpathyMercedes Drive PilotNHTSATeslaWaymo