Introduction
LiDAR has become the backbone of 3D perception for Physical AI, mobile robotics, autonomous vehicles, and Industry 4.0. But the market has fundamentally changed: five years ago, mechanical spinning dominated unchallenged. Today, solid-state technologies (MEMS, flash/SPAD, OPA) represent over 60% of shipments, driven by automotive and indoor robotics.
Each architecture addresses a different trade-off between range, FoV, resolution, reliability and cost. This comparison details the four major families with objective criteria, product benchmarks and a decision guide to choose the right sensor.
1. Mechanical Spinning
Principle: A complete optical module (multiple laser + photodetector pairs) is mounted on a rotor spinning at constant speed (5-20 Hz). Each laser channel illuminates a vertical slice; rotation sweeps 360°. Direct ToF measures distance. Flagship products: Ouster OS0/OS1/OS2, Hesai OT128, QT128, Pandar128.
Advantages: 360° FoV, TRL 9, high density (up to 26 M pts/s), range 200-300 m @10%, native ROS2, large ecosystem.
Disadvantages: Moving parts (motor + bearings) → lifespan 5,000-15,000 h. Bulky (H 80-150 mm, diameter 100-180 mm). Cost $1,500-8,000. Power 12-25 W. Vibration-sensitive.
Products:
| Model | Channels | FoV H/V | Range @10% | Price | ROS2 |
|---|---|---|---|---|---|
| Ouster OS2-128 REV8 | 128 | 360°×22.5° | 300 m | ~$6,000 | Yes |
| Ouster OS1-64 REV8 | 64 | 360°×45° | 150 m | ~$3,500 | Yes |
| Hesai OT128 | 128 | 360°×70° | 230 m | ~$3,500 | Yes |
| Hesai QT128 | 128 | 360°×105° | 150 m | ~$4,000 | Yes |
| Hesai Pandar128 | 128 | 360°×40° | 300 m | ~$8,000 | Yes |
2. MEMS (Semi-Solid-State)
Principle: A micro-machined silicon mirror (few mm) is electrostatically actuated to steer a laser beam along two axes. Single moving part <1 g. Products: Livox Mid-360 (rotating base + MEMS), InnovizTwo, RoboSense M1, Valeo Scala 3.
Advantages: Compact (<500 g), 50 G robustness, range 150-300 m @10%, cost $300-1,500, TRL 8-9 (BMW, Volvo deployment).
Disadvantages: Limited FoV (90-120° H, 25-40° V), mirror lifespan 10,000-50,000 h, non-rectangular scan pattern (Lissajous), irregular density.
Products:
| Model | FoV H/V | Range @10% | Price | Feature |
|---|---|---|---|---|
| Livox Mid-360 | 360°×59° | 200 m | ~$1,500 | Rotating base + MEMS |
| InnovizTwo | 120°×40° | 300 m | ~$500 | ASIL-B, automotive |
| RoboSense M1 | 120°×25° | 200 m | ~$700 | Largest volume worldwide |
| Valeo Scala 3 | 120°×30° | 300 m | TBD | Mercedes series |
3. Flash / SPAD Array
Principle: A VCSEL flood-illuminates the entire scene with one wide pulse. A SPAD/CMOS array captures the return. Each pixel measures ToF. No mechanical scanning. Products: Blickfeld Qb2, Hesai AT128/FT768, Ouster REV8 OS0/OS1.
Advantages: Zero moving parts (>100,000 h), compact (<100 cm³), low latency (10-30 Hz), potential cost <$200 in volume, perfect shock robustness.
Disadvantages: Range 50-150 m @10%, FoV 60-120°, limited resolution (VGA to 1 MP), solar noise sensitivity, dynamic range issue bright/dark objects.
Products:
| Model | Resolution | FoV H/V | Range @10% | Price | Target |
|---|---|---|---|---|---|
| Blickfeld Qb2 | QVGA (320×240) | 120°×80° | 120 m | ~$1,000 | Robotics, building |
| Hesai FT768 | 768×576 | 120°×90° | 150 m | ~$2,000 | Automotive |
| Hesai AT128 | 128×288 | 120°×25.4° | 200 m | ~$500 | Series auto (4M/yr) |
| Ouster REV8 OS0 | 128×128 | 90°×90° | 50 m | ~$2,000 | Indoor robotics |
| Ouster REV8 OS1 | 128×128 | 45°×45° | 150 m | ~$3,500 | Outdoor robotics |
4. OPA (Optical Phased Array)
Principle: A nano-antenna array on a silicon photonic chip steers the beam via electronic phase shifting. Purely electronic scanning, no moving parts. Analogy: optical AESA radar.
Potential advantages: Zero moving parts, MHz scanning, CMOS foundry production, potential cost <$100, programmable resolution, multi-beam.
Current limitations: TRL 4-6 (lab prototypes), range 10-50 m, fabrication yield <10%, FoV 30-60°, insufficient optical power for automotive.
Players: Quanergy (bankrupt 2022), Analog Photonics (USA defense), Point2 Technology (Korea). No commercial product. Not viable in 2026.
5. Comparison Table
| Criterion | Spinning | MEMS | Flash/SPAD | OPA |
|---|---|---|---|---|
| Horizontal FoV | 360° | 90-120° | 60-120° | 30-60° |
| Vertical FoV | 15-105° | 25-40° | 25-90° | 10-30° |
| Range @10% | 150-300 m | 150-300 m | 50-150 m | 10-50 m |
| Angular resolution | 0.05-0.2° | 0.1-0.3° | 0.2-0.5° | 0.1-0.5° (prog.) |
| Points/s | 1-26 M | 100 K-1 M | 100 K-5 M | 100 K-10 M (pot.) |
| Moving parts | Motor+bearings | 1 micro-mirror | None | None |
| Lifespan | 5,000-15,000 h | 10,000-50,000 h | >100,000 h | >100,000 h (th.) |
| Shock robustness | Medium | Good (50 G) | Excellent | Excellent |
| Cost (2026) | $1,500-8,000 | $300-1,500 | $200-2,000 | <$100 (pot.) |
| TRL | 9 | 8-9 | 7-8 | 4-6 |
| ROS2 support | Native | Partial | Partial | No |
| Power consumption | 12-25 W | 8-15 W | 5-15 W | <5 W (pot.) |
| Weight | 500-1,500 g | 200-500 g | 50-400 g | <50 g (pot.) |
Interpretation: Spinning dominates on 360° FoV and density but is expensive and wears. MEMS is the best cost/range/compactness trade-off. Flash wins on robustness and cost but loses on range. OPA is not viable in 2026.
6. Recommended Use Cases
Spinning → Mobile mapping (MMS), surveying, large-environment SLAM, tunnel/road/railway inspection, 360° perimeter surveillance.
MEMS → AMR/AGV, autonomous vehicles level 2+-4, delivery drones, humanoid robots, precision agriculture.
Flash/SPAD → Indoor robotics (warehouses, logistics), cobotics, smart building, automotive ADAS (AT128 BYD/Li Auto), fail-safe applications.
2D Safety LiDAR → Certified 2D scanners (SICK, Keyence, Hokuyo) IEC 61496 SIL2/PLd on industrial AMRs. Complementary to 3D.
7. Decision Guide
Mobile robot / AMR → MEMS (outdoor) or flash/SPAD (indoor). MEMS: Livox Mid-360, RoboSense M1. Flash: Ouster REV8 OS0, Blickfeld Qb2.
Mapping → Mechanical spinning. Hesai Pandar128 or Ouster OS2-128 for cm-level SLAM. Drone: Hesai QT128, Ouster OS0-128.
Series automotive → Flash (Hesai AT128, 4M units/yr, BYD/Li Auto/GAC) or MEMS (RoboSense M1, InnovizTwo). Cost <$500, -40/+85°C, ASIL.
Indoor → Flash/SPAD. 50 m sufficient, no moving parts = unlimited reliability, no solar noise = excellent SNR.
Safety → Certified 2D spinning (SICK, Keyence) SIL2/PLd mandatory on AMRs Europe (Machinery Directive 2006/42/EC).
Humanoid robot → MEMS and flash (weight 200-450 g). Spinning too heavy. Figure, Tesla Optimus, Unitree use MEMS + flash.
8. 2026-2028 Trends
- Solid-state convergence: spinning retreats in automotive and consumer robotics. Solid-state (MEMS + flash) grows 35%/year.
- Cost decline: -40 to 60% between 2024 and 2026. AT128 ~$500, InnovizTwo ~$500 in volume. Spinning >$2,000.
- SoC LiDAR: Hesai Picasso SPAD-SoC, Ouster REV8 native color on-chip. Native color becomes standard in 2026.
- Premium FMCW: Aeva (Aeries II, Daimler Truck), Mobileye. Instant velocity measurement. Cost >$1,000, TRL 6-7.
- 2D safety: remains essential. IEC 61496 certification long and costly for 3D sensors.
Conclusion
There is no "best" LiDAR — only the right sensor for your application. Spinning is unbeatable for mapping. MEMS offers the best trade-off. Flash is the robust, economical choice. OPA is a 2028+ promise.
Use the Jovaxis comparator — filter by technology, range, cost or application. Find the LiDAR suited to your project, compare datasheets, access data verified by our experts.
China
United States
Germany