No. Tesla Optimus does not use LiDAR — and it never has. The robot perceives the world with eight Autopilot-grade cameras plus proprioceptive and tactile sensing, mirroring the camera-only bet Tesla made on its cars. Here is the complete, honestly-sourced sensor stack, the camera count, and what is actually known about water resistance.
👉 The single most useful thing to know: on humanoid spec pages, sensors are the most frequently fabricated section. Tesla publishes far less hardware detail than aggregator sites imply. This guide flags what is confirmed versus what is inferred.
- LiDAR: Not used. No LiDAR, no radar, no depth-scanner module. Vision-only, by deliberate design and cost logic.
- Cameras: 8 Autopilot-grade cameras providing 360° coverage and stereo depth estimation. Tesla has never published a full camera-placement map.
- Other sensing: IMU, joint torque sensors in every actuator, foot force/torque sensors, fingertip tactile sensors, microphones and speakers.
- Water resistance: Tesla has published no official IP rating. The IP54 and IP68 figures circulating online come from third-party spec aggregators, not from Tesla.
- Suppliers: Roughly 70% of Optimus components trace to Chinese suppliers — Sanhua, Tuopu, Leaderdrive and Wuzhou Xinchun lead the actuator and reducer chain.
Does Tesla Optimus Use LiDAR? The Direct Answer
Optimus uses a pure vision system with no LiDAR and no radar: eight Autopilot cameras for 360° awareness, stereo depth estimation, real-time mapping, and force/torque sensing for balance (BotInfo.ai technical analysis). This is not an oversight or a cost-cutting compromise applied late — it is the same first-principles position Tesla has defended since it stripped radar from its vehicles in 2021 and ultrasonic sensors in 2022.
Why Tesla refuses LiDAR on Optimus
Tesla’s argument runs on three legs. First, cost: LiDAR units are orders of magnitude more expensive than camera modules, which matters enormously when the stated goal is a $20,000 robot at scale. Second, architecture reuse: Optimus inherits the FSD neural-network stack trained on billions of real-world driving frames. Third, philosophy — Tesla frames the challenge as "not a sensor problem, it’s an AI problem." (vision-only strategy breakdown)
Tesla Autopilot director Ashok Elluswamy reinforced this publicly on April 27, 2026, posting a demonstration video captioned "The advantage of having eyes all around." (Basenor). For a deeper account of how Tesla arrived at this position, see this analysis of Tesla’s radar and LiDAR exit.
The counter-argument — and where it actually bites indoors
The rest of the industry disagrees. Rivian, adding LiDAR to vehicles in 2026, argues cameras are passive light sensors that underperform in darkness and fog (The Drive). For a home or factory humanoid, though, the trade-offs shift. Weather is irrelevant indoors. Distance ranging matters less at 1–3 meters than at 100 meters.
Three indoor failure modes are where the missing LiDAR genuinely costs Optimus something:
- Low light. A dark hallway or an unlit basement degrades passive cameras in a way an active sensor would not.
- Textureless surfaces. Blank white walls, plain countertops and monochrome floors give stereo vision almost nothing to triangulate against.
- Thin and reflective obstacles. Chair legs, cables, mirrors. Worth noting: LiDAR is also weak on glass and mirrors, so this one is not a clean win for either camp.
💡 Nuance most articles miss: for a humanoid, the vision-versus-LiDAR argument is less decisive than the tactile-sensing argument. Once a robot is within arm’s reach of an object, touch and torque feedback do more work than any ranging sensor. Tesla’s Gen 3 leap was in the hands, not the eyes.
Tesla Optimus Sensors: The Complete 2026 Stack
The table below separates confirmed sensing from disputed claims. Full hardware context lives in our Optimus hardware and actuator breakdown.
| Sensor type | Quantity / spec | Function | Confidence |
|---|---|---|---|
| Autopilot cameras | 8 | 360° vision, stereo depth, object recognition, mapping | High — consistently reported across all sources |
| IMU (inertial unit) | 1 (torso) | Balance, orientation, fall detection, gait stabilization | High |
| Joint torque sensors | In every actuator | Force feedback, collision detection, compliant motion | High |
| Foot force/torque sensors | 2-axis feet | Dynamic balance, slip detection, terrain adaptation | High |
| Fingertip tactile sensors | All fingers; 22 DoF hands (Gen 3) | Grip-pressure modulation — the egg-handling demo | High |
| Dual encoders per joint | Input + output position | Sub-degree precision control | Medium-high — engineering reports |
| Microphones / speakers | Not specified | Voice commands via Grok integration, audio feedback | Medium |
| Ultrasonic proximity | Listed by some aggregators | Close-range obstacle detection | Low — likely carried over from car specs |
| LiDAR / radar | None | — | Confirmed absent |
How many cameras does Tesla Optimus have?
Eight. That figure is consistent across every credible technical source, and it matches the Autopilot camera count on Tesla vehicles — which is the point, since the perception stack is shared. What Tesla has never published is a precise placement diagram: how many sit in the head, whether any are torso or hip-mounted, focal lengths, or resolution.
Treat any site that gives you exact per-camera specs with suspicion. As of August 2026 that detail is not public, and the full V3 body specification has not been formally released.
The sensing that actually determines usefulness
The Gen 3 hands doubled degrees of freedom from 11 to 22 and increased actuator count 4.5× to 50 total, with all 25 actuators per side moved into the forearm in a tendon-driven, biomimetic layout (Gen 3 hands breakdown). Every fingertip carries force-feedback hardware. That is what lets the robot pick up an egg without crushing it — not the cameras.
✔ Buying heuristic for any humanoid in 2026: count degrees of freedom per hand and ask whether fingertips have force feedback. A robot with 30 cameras and rigid grippers is less useful in a home than one with 8 cameras and 22-DoF tactile hands.
Tesla Optimus Water Resistance and IP Rating: The Honest Answer
This is the most misreported specification in the entire Optimus stack, so let us be precise: Tesla has not published an official IP rating for any Optimus generation. Every number you see quoted is a third-party estimate or an aggregator’s inference.
What is actually circulating
- IP54 (Gen 2): appears on several spec-aggregator pages alongside actuator cooling figures. No Tesla source confirms it.
- IP68 for arms and end-effectors (Gen 3): listed by robot-spec databases as a partial, localized rating rather than a whole-body one (RobotHub spec sheet).
- "Water-resistant hands": widely repeated on YouTube and content farms with no primary source attached.
What IP ratings mean before you trust one
The IEC 60529 code has two digits: solids first, liquids second. The difference between the numbers being thrown around is not cosmetic.
| Rating | What it protects against | What it means for household use |
|---|---|---|
| IP54 | Limited dust ingress; splashing water from any direction | Survives a splash and kitchen humidity. Not safe in a filled sink. |
| IP65 | Dust-tight; low-pressure water jets | Can be sprayed down. Adequate for wet-floor cleaning tasks. |
| IP67 | Dust-tight; immersion to 1 m for 30 minutes | Hands could be submerged in a sink briefly. |
| IP68 | Dust-tight; continuous immersion beyond 1 m | True dishwashing-in-the-sink capability, sustained. |
Why whole-body IP68 is implausible — and localized sealing is not
An engineering reality check that the spec sheets skip. Optimus carries a roughly 2.3 kWh torso battery, 28 structural actuators, and compute that requires active thermal management. Cooling paths mean airflow, and airflow means openings. Sealing an entire 57 kg biped to continuous-immersion standard while dissipating that heat is not a realistic 2026 engineering outcome.
Sealing the forearms and hands to a high rating, however, is entirely plausible — and it is exactly what the Gen 3 architecture enables, because all 25 actuators per side were relocated into the forearm and driven through tendons. A sealed tendon pass-through is a well-understood mechanical problem. That is the design decision behind the dishwashing claims.
Tesla Board Chair Robyn Denholm has stated the Gen 3 hands can "already wash clothes, clean dishes, and shake hands" with sensitive touch (reported analysis of the Gen 3 hands). Read that carefully: it is a claim about hand capability, not a certified ingress-protection rating.
⚠ Practical translation for 2026: assume splash-resistant hands, not a waterproof robot. Do not expect an early Optimus to work in rain, mop a flooded floor, or be hosed down. Any listing that advertises a specific IP number as "official Tesla spec" is reselling an aggregator guess.
Optimus Actuators, Motors and Sensors: Who Actually Builds Them
Tesla designs the architecture and outsources most component manufacturing. Supply-chain research from Chinese securities firms indicates Tesla completed factory audits for the third-generation robot with Chinese suppliers holding roughly 70% of component share (BigGo Finance supply-chain report).
| Supplier | Component | Position in the chain |
|---|---|---|
| Tuopu Group | Linear + rotary actuator assembly | Tier 0.5; Tesla is 35–40% of revenue; Mexico plant already producing |
| Sanhua Intelligent Controls | Joint modules, actuators, thermal control | Core assembly supplier; Mexico deliveries from 2026 |
| Leaderdrive / Green Harmonic | Harmonic reducers | Passed Tesla supply-chain verification; 60%+ domestic share |
| Wuzhou Xinchun | Planetary roller screws | High-torque hip and knee joints; Thailand plant online 2026 |
| Xinjian, Beite, Zhenyu | Roller screws, full linear actuator modules | Mechanical stack — converts electrical energy into force |
| Keli Sensing, Zhaowei | Force sensors, precision micro-gearboxes | Sensing and fine-motion layer |
| TSMC / Samsung | AI5 inference chip | The only critical component not sourced from China |
Two structural risks sit inside that table. Actuators account for roughly half the robot’s component value, and each unit needs several kilograms of NdFeB rare-earth magnets — a supply China dominates. Near-shoring to Thailand and Mexico is underway precisely because of that exposure (36Kr supply-chain analysis). Our full breakdown: who makes Tesla Optimus parts.
👉 Investor-relevant read: the sensor debate is cheap talk, but the actuator chain is where the $20K price target lives or dies. Planetary roller screws alone — 14 units per robot at $1,350–$2,700 each by Morgan Stanley estimates — can exceed the entire target retail price at current pricing.
Quick Comparison: Optimus Sensing vs. Rival Humanoids
| Tesla Optimus | Figure 03 | Boston Dynamics Atlas (electric) | |
|---|---|---|---|
| LiDAR | No | No (vision + onboard sensing) | Uses depth sensing |
| Cameras | 8 Autopilot-grade | Multi-camera head array | Multi-modal perception suite |
| Hand DoF | 22 per hand (Gen 3) | 16 per hand (prior gen) | Not publicly specified; 56 DoF whole-body |
| Tactile sensing | Fingertip force feedback | Sensors in each finger | Finger and palm tactile |
| Official IP rating | Not published | Not published | Not published |
| Design priority | Mass-manufacturability and cost | Dexterity + commercial pilots | Industrial durability and payload |
Worth noting: no major humanoid manufacturer publishes a certified IP rating. This is an industry-wide disclosure gap, not a Tesla-specific one. For the wider capability picture, see our Optimus capabilities guide.
Checklist: How to Verify Any Optimus Spec Claim
Optimus is the single most spec-fabricated product in consumer robotics. Run any claim through these five filters before repeating it.
- Trace it to a primary source: a Tesla earnings call, an official Tesla post, or an executive on the record. Aggregator pages citing other aggregators are not sources.
- Check whether the claim describes Gen 2, "Gen 3" (hands only), or V3 (full new body). Most spec confusion is a generation mix-up.
- Reject any precise number Tesla has never published — IP ratings, camera resolutions, exact battery chemistry, joint-level torque figures.
- Ask whether the spec is physically consistent with the rest of the robot. Whole-body waterproofing and active cooling are in tension.
- Prefer supply-chain disclosures over marketing. What a supplier tells its shareholders is more reliable than what a spec page tells you.
💡 The V3 full-body reveal is when most of these open questions get answers. Until Tesla publishes a formal spec sheet, "not disclosed" is the correct answer to a lot of Optimus hardware questions — and saying so is a mark of accuracy, not ignorance.
FAQ
Does Tesla Optimus use LiDAR or radar?
No. Optimus is vision-only, using eight Autopilot cameras plus inertial, torque and tactile sensing. Tesla has excluded LiDAR and radar deliberately, on cost and architecture grounds.
How many cameras does Tesla Optimus have?
Eight Autopilot-grade cameras providing roughly 360° coverage with stereo depth estimation. Tesla has not released placement diagrams or per-camera specifications.
What is the Tesla Optimus IP rating?
There is no official one. IP54 and IP68 figures circulate widely but originate from third-party spec databases. Localized sealing of the forearms and hands is plausible; whole-body waterproofing is not.
Can Optimus wash dishes or work around water?
Tesla’s board chair has said the Gen 3 hands can wash clothes and clean dishes. Treat that as a dexterity claim. Without a published ingress rating, assume splash resistance only.
Who supplies Optimus actuators, motors and sensors?
Primarily Chinese manufacturers — Tuopu and Sanhua for actuators and joint modules, Leaderdrive for harmonic reducers, Wuzhou Xinchun and Zhenyu for roller screws. The AI5 chip is fabricated by TSMC and Samsung.
Would adding LiDAR make Optimus better at home?
Marginally, in dark rooms and against textureless surfaces. It would not solve glass, mirrors or fine manipulation — which are the failure modes that actually matter indoors.
The Bottom Line
Optimus is a vision-only humanoid: eight cameras, no LiDAR, no radar, with the real perceptual sophistication concentrated in torque and tactile feedback rather than ranging sensors. That is a defensible engineering position for an indoor robot, and a genuinely risky one in low light.
On water resistance, the accurate answer in August 2026 is that Tesla has not told us. Sealed forearms are architecturally plausible and consistent with the dishwashing claims. A waterproof robot is not.
And the spec sheet you should actually be watching is the supplier chain, not the sensor list — that is where the $20,000 target is won or lost. We update this page at every hardware disclosure; track the current generation in our latest Optimus version guide and the Gen 3 technical tracker.
💡 Want the V3 spec sheet the hour Tesla publishes it — with the aggregator noise filtered out? Join the Optimus Investors Club for verified hardware updates only.
Sources: BotInfo.ai, Basenor, Not a Tesla App, The Drive, RobotHub, BigGo Finance, 36Kr, RobotToday, Tesla Q1–Q2 2026 earnings calls. Last fact-check: August 5, 2026.
NEVER MISS AN OPTIMUS UPDATE
We track every Tesla Optimus development — specs, deployment milestones, pricing and competitive moves — updated as news breaks.