Solid-State LiDAR and Photonic Integrated Circuits for Airside Autonomous GSE

Introduction & Motivation
LiDAR Beam Steering Technologies
FMCW vs ToF: Measurement Principles
Silicon Photonics Integration
Solid-State LiDAR Products and Specifications
Optical Phased Arrays (OPA)
MEMS Mirror LiDAR
Flash LiDAR
Airside-Specific Requirements
Reliability and Lifetime Analysis
Integration with reference airside AV stack Stack
Cost Roadmap and Market Dynamics
Migration Strategy: Mechanical to Solid-State
Implementation Roadmap
Key Takeaways
References

1. Introduction & Motivation

1.1 The Mechanical LiDAR Problem for 24/7 Airside Operations

the reference airside AV stack's current fleet uses 4-8 RoboSense mechanical/semi-mechanical LiDARs per vehicle (RSHELIOS, RSBP). These sensors rely on spinning optical assemblies or MEMS mirrors with moving parts that face accelerated wear in the harsh airside environment:

Stress Factor	Impact on Mechanical LiDAR	Airside Severity
Vibration	Bearing wear, optical misalignment	High — tarmac surface, loading/unloading
Temperature cycling	Thermal expansion of bearings, lubricant degradation	Extreme — -20°C to +60°C tarmac
De-icing chemicals	Seal degradation, ingress	Severe — glycol/potassium formate spray
Jet blast	Mechanical shock, debris impact	Severe — 100+ km/h gusts at gate
Dust/particulate	Abrasion of optical window, bearing contamination	Moderate — rubber particles, sand
24/7 duty cycle	16-20 hours/day continuous rotation	Very high — 5,800-7,300 hours/year

Mechanical LiDAR MTBF in airside conditions: Estimated 15,000-25,000 hours (2-3.5 years), with degraded performance well before failure. For a 50-vehicle fleet with 6 LiDARs each, this means ~100-200 LiDAR replacements per year at $2,000-5,000 each = $200K-1M/year in sensor replacement alone.

1.2 The Solid-State Promise

Solid-state LiDAR eliminates all moving parts, offering:

10-100x longer MTBF: 100,000+ hours projected (semiconductor-level reliability)
Smaller form factor: Matchbox-sized sensors vs. current hockey-puck mechanical units
Lower power: 5-15W vs. 15-30W for mechanical spinning LiDAR
Instant-on: No spin-up time (mechanical LiDARs need 2-5 seconds)
Per-point velocity: FMCW solid-state provides instantaneous radial velocity on every measurement
Immunity to solar interference: Coherent detection in FMCW inherently rejects ambient light

1.3 Technology Readiness

As of early 2026, solid-state LiDAR technologies are at varying TRL levels:

Technology	TRL	Status	First Mass Production
MEMS mirror	TRL 8-9	In production vehicles	2022 (Livox, Innoviz)
Flash LiDAR	TRL 7-8	Automotive qualification	2024-2025 (Ibeo, Continental)
OPA (silicon photonics)	TRL 6-7	Pre-production prototypes	2026-2028 (Voyant, Analog Photonics)
FMCW + OPA	TRL 5-7	Advanced demos, early products	2026-2028 (Aeva, SiLC)
Photonic focal plane array	TRL 5-6	First prototypes	2027-2029 (Voyant Helium)

2. LiDAR Beam Steering Technologies

2.1 Technology Comparison

┌─────────────────────────────────────────────────────────────────┐
│              LIDAR BEAM STEERING TAXONOMY                       │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  MECHANICAL              SEMI-SOLID-STATE      FULLY SOLID-STATE│
│  ┌──────────┐           ┌──────────┐          ┌──────────────┐ │
│  │ Spinning │           │  MEMS    │          │    OPA        │ │
│  │ Mirror   │           │  Mirror  │          │ (Phased Array)│ │
│  │          │           │          │          │               │ │
│  │ Velodyne │           │ Innoviz  │          │ Voyant        │ │
│  │ Ouster   │           │ Livox    │          │ Analog Photon │ │
│  │ RoboSense│           │ MicroVis │          │               │ │
│  └──────────┘           └──────────┘          ├──────────────┤ │
│                                                │   Flash       │ │
│  Moving: rotor/motor    Moving: MEMS mirror    │ (Illuminator) │ │
│  Lifetime: 15-30K hrs   Lifetime: 30-50K hrs   │               │ │
│  Power: 15-30W          Power: 8-20W           │ Ibeo          │ │
│                                                │ Continental   │ │
│                                                ├──────────────┤ │
│                                                │ Focal Plane   │ │
│                                                │ Array (FPA)   │ │
│                                                │               │ │
│                                                │ Voyant Helium │ │
│                                                │               │ │
│                                                │ No moving     │ │
│                                                │ parts at all  │ │
│                                                │ Lifetime:     │ │
│                                                │ 100K+ hrs     │ │
│                                                │ Power: 3-10W  │ │
│                                                └──────────────┘ │
└─────────────────────────────────────────────────────────────────┘

2.2 Beam Steering Methods Compared

Method	Moving Parts	Scan Rate	FoV	Angular Res	Power	Cost Trend
Spinning mirror	Motor + mirror assembly	10-20 Hz	360° × 40°	0.1-0.4°	15-30W	Flat ($500-2000)
MEMS mirror	Micro-mirror (1-3mm)	5-30 Hz	120° × 25°	0.05-0.2°	8-20W	Declining ($300-1000)
OPA	None (electronic)	~GHz point-to-point	50-120° × 20°	0.01-0.05°	3-10W	Steep decline (→$100)
Flash	None (flood illuminator)	Frame rate (10-30 Hz)	20-120° × 10-30°	0.1-0.5°	5-15W	Moderate ($200-800)
Focal plane array	None (2D on-chip steering)	Electronic	Up to 180°	0.01-0.1°	3-8W	Steep decline (→$50-200)

2.3 Scan Patterns

Mechanical spinning: Continuous 360° rotation, uniform angular sampling. Well-suited for current reference airside AV stack (360° coverage from fewer sensors).

MEMS raster: Lissajous or raster patterns. Non-uniform point density with higher density at edges (mechanical turnaround). Some designs (Livox, Innoviz) use optimized patterns for more uniform coverage.

OPA random access: Electronic steering enables arbitrary point placement. Can dynamically allocate resolution — high density on nearby aircraft, sparse on empty taxiway. This is transformative for foveated perception strategies (see active-perception-sensor-scheduling.md).

Flash: Entire FoV illuminated simultaneously. Uniform point density but limited range (lower energy per point). Best for short-range, wide-FoV applications (docking, close-range obstacle detection).

3. FMCW vs ToF: Measurement Principles

3.1 Time-of-Flight (Current reference airside AV stack Approach)

Pulsed ToF:
                ┌──┐                          ┌──┐
  Transmit:     │  │                          │  │
  ──────────────┘  └──────────────────────────┘  └──────
                                                        
                         ┌──┐                          ┌──┐
  Receive:               │  │                          │  │
  ───────────────────────┘  └──────────────────────────┘  └──
                         
                ├────────┤
                  Δt = 2R/c
                  
  Range R = c·Δt/2
  Precision: ~1-5 cm (limited by pulse width and timing jitter)
  Velocity: NOT measured (requires multi-return tracking)

Limitations for airside:

No per-point velocity → MOT tracker required for dynamic objects
Vulnerable to solar interference (905nm near-IR competes with sunlight)
Range-energy tradeoff: longer range needs higher peak power (eye safety limit)
Multi-echo interference from rain, fog, de-icing spray

3.2 Frequency-Modulated Continuous Wave (FMCW)

FMCW LiDAR:
                  Frequency
                    ↑    /\    /\    /\
  Transmit (TX):    │   /  \  /  \  /  \
                    │  /    \/    \/    \
                    │ /                  \
                    └──────────────────────→ Time
                    
                    ↑      /\    /\    /\
  Receive (RX):     │     /  \  /  \  /
  (delayed+shifted) │    /    \/    \/
                    │   /
                    └──────────────────────→ Time
                    
  Beat frequency: f_beat = f_range + f_doppler
  
  Upsweep:   f_beat_up   = (2·B·R)/(c·T) + (2·v·f_0)/c
  Downsweep: f_beat_down = (2·B·R)/(c·T) - (2·v·f_0)/c
  
  Range:    R = c·T·(f_up + f_down)/(4·B)
  Velocity: v = c·(f_up - f_down)/(4·f_0)
  
  Where: B = chirp bandwidth, T = chirp period, f_0 = center frequency

FMCW advantages for airside:

Capability	Benefit for Airside
Per-point instantaneous velocity	Detect approaching aircraft (1-5 km/h pushback), GSE, personnel without tracking delay
Coherent detection	Immune to solar interference — critical for open-air tarmac operations
Lower peak power	10-100mW CW vs. 10-100W pulsed → inherently eye-safe at Class 1
Range precision	mm-level (vs cm-level for ToF) — better for precision docking
Interference rejection	Heterodyne detection rejects other LiDAR signals, ambient light, headlights
Weather resilience	Coherent gain suppresses multi-path scattering in fog/rain by ~10-20 dB

FMCW challenges:

Challenge	Current State
Point rate	100K-1M points/s (vs. 1-3M for mechanical ToF) — improving rapidly
Complexity	Requires stable laser source, balanced photodetectors, DSP — silicon photonics solves this
Range	100-300m demonstrated (vs. 200-300m mechanical ToF) — sufficient for airside (max 200m needed)
Cost	Currently 2-5x ToF — rapidly converging with silicon photonics volume

3.3 Velocity Measurement Value

For airside operations, per-point velocity is transformative:

python

# Current approach: velocity requires multi-frame tracking
# Problem: latency of 3-5 frames (300-500ms at 10 Hz) to establish velocity

# FMCW approach: instantaneous velocity per point
class FMCWPointCloud:
    """Point cloud with per-point radial velocity from FMCW LiDAR."""
    
    def __init__(self):
        self.x = []      # meters
        self.y = []      # meters  
        self.z = []      # meters
        self.intensity = []  # reflectivity
        self.velocity = []   # m/s radial velocity (positive = approaching)
    
    def detect_approaching_objects(self, threshold_velocity=-0.5):
        """
        Instantly identify approaching objects without tracking.
        
        In current ToF pipeline: requires 3-5 frames of MOT tracking.
        With FMCW: single scan, zero-latency velocity.
        """
        approaching_mask = np.array(self.velocity) < threshold_velocity
        approaching_points = self.get_points(approaching_mask)
        
        # Cluster approaching points
        clusters = DBSCAN(eps=0.5, min_samples=5).fit(approaching_points[:, :3])
        
        objects = []
        for label in set(clusters.labels_) - {-1}:
            cluster_mask = clusters.labels_ == label
            cluster_points = approaching_points[cluster_mask]
            objects.append({
                'centroid': cluster_points[:, :3].mean(axis=0),
                'velocity': cluster_points[:, 4].mean(),  # mean radial velocity
                'num_points': cluster_mask.sum(),
                'extent': cluster_points[:, :3].ptp(axis=0),  # bounding box
            })
        
        return objects
    
    def jet_blast_boundary_detection(self):
        """
        Detect jet blast boundaries via velocity field.
        
        FMCW can detect particle velocity in jet exhaust plume.
        This is invisible to ToF LiDAR, camera, and radar.
        Thermal cameras detect temperature but not flow velocity.
        
        FMCW LiDAR is the ONLY sensor that can detect both
        the spatial extent AND flow velocity of jet blast.
        """
        # High-velocity points in exhaust direction behind aircraft
        # Jet blast particles (dust, water droplets) have measurable velocity
        high_velocity_mask = np.abs(np.array(self.velocity)) > 5.0  # >5 m/s
        
        if high_velocity_mask.sum() > 50:
            blast_points = self.get_points(high_velocity_mask)
            # Fit boundary surface to high-velocity region
            boundary = convex_hull_2d(blast_points[:, :2])
            return JetBlastZone(boundary, max_velocity=np.max(np.abs(blast_points[:, 4])))
        
        return None

4. Silicon Photonics Integration

4.1 What is Silicon Photonics?

Silicon photonics (SiPh) fabricates optical components — waveguides, modulators, photodetectors, couplers — on standard silicon wafers using CMOS-compatible processes. This enables:

Mass production: Existing semiconductor fabs (TSMC, GlobalFoundries, TowerSemiconductor) can produce photonic chips at automotive volumes
Monolithic integration: Laser source, beam steering, receiver, and signal processing on a single chip
Cost reduction: From $1,000+ per discrete-optics LiDAR to projected $50-200 at volume
Size reduction: Entire LiDAR transceiver in <1 cm² die area

4.2 Key Components on Silicon

┌─────────────────────────────────────────────────────┐
│           SILICON PHOTONICS LiDAR-ON-CHIP           │
│                                                     │
│  ┌─────────┐   ┌──────────┐   ┌───────────────┐   │
│  │ Laser   │──→│ Frequency│──→│ Optical Phased │──→ FREE SPACE
│  │ Source   │   │ Chirp    │   │ Array (OPA)    │   │  (TO TARGET)
│  │ (III-V  │   │ Modulator│   │ 128-8192       │   │
│  │ bonded) │   │          │   │ emitters       │   │
│  └─────────┘   └──────────┘   └───────────────┘   │
│       │                            ↑                │
│       │         ┌──────────┐       │                │
│       └────────→│ Reference│   ┌───┴──────────┐    │
│                 │ Splitter │   │ Phase Shifters│    │
│                 └─────┬────┘   │ (thermo-optic │    │
│                       │        │  or PN-junction│    │
│                       ↓        │  per channel) │    │
│                 ┌──────────┐   └───────────────┘    │
│                 │ Balanced  │                        │
│  FREE SPACE ──→│ Photo-    │──→ Electrical beat     │
│  (FROM TARGET)  │ detectors │    frequency signal   │
│                 │ (Ge/Si)   │                        │
│                 └──────────┘                        │
│                       │                              │
│                       ↓                              │
│                 ┌──────────┐                        │
│                 │ ADC +    │──→ Range + Velocity    │
│                 │ DSP      │    per point           │
│                 └──────────┘                        │
└─────────────────────────────────────────────────────┘

4.3 Material Platforms

Platform	Waveguide Loss	Advantage	Limitation	Vendors
SOI (Silicon on Insulator)	1-2 dB/cm	CMOS compatible, high index contrast	No native laser, limited to 1.1-4 μm	Most SiPh fabs
SiN (Silicon Nitride)	0.01-0.1 dB/cm	Ultra-low loss, wider transparency	Lower index contrast, larger waveguides	LioniX, Ligentec
III-V on Si (hybrid)	0.5-2 dB/cm	Native laser, photodetector	Complex bonding, lower yield	Aeva, Intel
InP	0.5-1 dB/cm	Full photonic integration	Not CMOS compatible, expensive	Lumentum, II-VI

For LiDAR: Hybrid III-V/Si or III-V/SiN approaches dominate because they combine the active gain medium (laser) from III-V semiconductors with the low-loss passive routing of silicon or silicon nitride waveguides.

4.4 Integration Levels

Level 1: Discrete optics (current)
  Laser → fiber → lens → mirror → target → lens → fiber → detector
  Components: 20-50 discrete parts
  Assembly: manual alignment, hours per unit
  Cost: $1,000-5,000

Level 2: Hybrid photonic module
  Laser chip → wire-bond → SiPh chip (waveguides + PDs) → package
  Components: 5-10 die + package
  Assembly: automated pick-and-place, minutes per unit
  Cost: $200-800 (Aeva Atlas, SiLC Eyeonic)

Level 3: Monolithic SiPh LiDAR-on-chip
  Single die: laser + modulator + OPA + detector + DSP
  Components: 1 die + package
  Assembly: wafer-level, seconds per unit
  Cost: $50-200 at volume (Voyant Helium target)

Level 4: Fully integrated sensor SoC (future)
  LiDAR transceiver + point cloud processor + interface on single package
  Cost: $20-100 at high volume (2030+ projection)

5. Solid-State LiDAR Products and Specifications

5.1 FMCW Solid-State Products (2024-2026)

Product	Company	Beam Steering	Range	Points/s	FoV	Velocity	Form Factor	Price (est.)	Status
Atlas	Aeva	MEMS + FMCW	300m @10%	~2M	120° × 30°	±200 m/s per point	Automotive-grade	$500-800	Production 2025
Aeries II	Aeva	MEMS + FMCW	500m @10%	~2M	120° × 30°	±200 m/s per point	Premium	$1,000-2,000	Available H1 2026
Helium	Voyant	Photonic FPA (no moving parts)	100-200m (est.)	12K-100K pixels	Up to ~180°	Per-point	<150g, <50 cm³	$200-500 (target)	Prototype CES 2026
Eyeonic Vision	SiLC	MEMS + FMCW	50-1,250m	Variable	Multiple configs	Per-point + polarization	Industrial module	$2,000-5,000	Available now
Scantinel	Scantinel	MEMS + FMCW	200m	500K	120° × 30°	Per-point	Automotive	$500-1,500	Sampling 2025

5.2 Semi-Solid-State Products (In Production)

Product	Company	Beam Steering	Range	Points/s	FoV	Price (est.)	Status
HAP	Livox	Non-repetitive MEMS	450m @10%	240K	120° × 25°	$800-1,200	Production
Two (InnovizTwo)	Innoviz	MEMS	300m	14.4M eq	120° × 25°	$500-1,000	Production in BMW
AT128	Hesai	Hybrid solid-state	200m	1.53M	120° × 25.4°	$300-800	Mass production
MX	RoboSense	MEMS	200m	750K	120° × 25°	$500-1,000	Production

5.3 Comparison with Current reference airside AV stack Sensors

Parameter	RoboSense RSHELIOS (current)	Aeva Atlas (FMCW)	Voyant Helium (full SS)
Beam steering	Spinning	MEMS	Photonic FPA
Moving parts	Motor + mirror	MEMS mirror (~1mm)	None
Range	150m @10%	300m @10%	100-200m (est.)
Points/s	1.2M	~2M	12K-100K
FoV	360° × 32°	120° × 30°	Configurable
Per-point velocity	No	Yes	Yes
Power	~15W	~12W	~5W (est.)
Weight	~600g	~500g	<150g
Operating temp	-20 to 60°C	-40 to 85°C	-40 to 85°C (target)
MTBF (est.)	20,000 hrs	40,000+ hrs	100,000+ hrs
IP rating	IP67	IP67/69K	IP67 (target)
Price	$1,500-3,000	$500-800	$200-500 (volume)

6. Optical Phased Arrays (OPA)

6.1 Operating Principle

An OPA steers a laser beam by controlling the phase of light emitted from an array of closely spaced waveguide emitters. By varying the relative phase between emitters, the constructive interference direction changes, steering the beam without any moving parts.

OPA Beam Steering:

  Phase shifters:   φ₁  φ₂  φ₃  φ₄  φ₅  φ₆  φ₇  φ₈
                    │   │   │   │   │   │   │   │
  Emitters:         ▼   ▼   ▼   ▼   ▼   ▼   ▼   ▼
                    ○   ○   ○   ○   ○   ○   ○   ○
                    
  Wavefront (φ₁=φ₂=...=φ₈):     Wavefront (linear phase ramp):
  
    ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃               ╲ ╲ ╲ ╲ ╲ ╲ ╲ ╲
    ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃                ╲ ╲ ╲ ╲ ╲ ╲ ╲ ╲
    ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓                 ↘ ↘ ↘ ↘ ↘ ↘ ↘ ↘
    Straight ahead                   Steered to angle θ
    
  Steering angle: sin(θ) = λ·Δφ / (2π·d)
  Where: λ = wavelength, Δφ = phase increment between emitters, d = emitter spacing
  
  Max angle: θ_max = arcsin(λ / 2d)  (grating lobe limit)
  For d = λ/2: θ_max = 90° (hemisphere)

6.2 OPA Performance State-of-the-Art (2025-2026)

Parameter	2022 SOTA	2025 SOTA	Airside Requirement
Channel count	128	512-8192	256+ for adequate resolution
Steering range (azimuth)	±25°	±50-60°	±60° (3-4 sensors for 360°)
Steering range (elevation)	Fixed or ±5°	±10-15°	±15° for close-range objects
Angular resolution	0.1°	0.01-0.05°	<0.1° for personnel at 100m
Switching speed	~μs per point	~ns per point	<100 μs (non-critical)
Side lobe suppression	-10 to -15 dB	-15 to -25 dB	<-20 dB (avoid ghost detections)
Power efficiency	5-10% wall-plug	10-20%	Higher is better for battery GSE

6.3 2D Steering Approaches

Wavelength + OPA (λ-OPA):

OPA steers in one axis (azimuth)
Wavelength tuning steers in the other axis (elevation) via dispersive grating
Advantage: Only 1D phase control needed (simpler electronics)
Limitation: Requires tunable laser (~40nm range for ±10° elevation)
Used by: Aeva (early), academic demonstrations

2D OPA:

True 2D array of emitters with 2D phase control
Advantage: Full electronic 2D steering, no tunable laser needed
Limitation: N² phase shifters for N×N array (power, complexity)
Used by: Voyant Helium (focal plane array variant)

Switch + OPA:

Optical switch selects between multiple fixed OPA sub-arrays
Each sub-array covers a different angular sector
Advantage: Lower phase-shifter count, higher power efficiency
Used by: Intel (research), various academic groups

6.4 OPA Challenges and Solutions

Challenge	Impact	Solution	Status
Grating lobes	Ghost beams at wrong angles	Sub-wavelength emitter pitch (<775nm for 1550nm)	Achieved in SiN
Power efficiency	High insertion loss limits range	Optimized waveguide design, SiN platforms	15-20% demonstrated
Beam quality	Poor far-field pattern	Apodization, non-uniform spacing	Active research
Phase noise	Pointing jitter	Feedback calibration, thermal stabilization	Solved for slow scan
Crosstalk	Adjacent channel interference	Isolation trenches, differential drive	Improving
2D scaling	N² complexity	Hierarchical addressing, row-column drive	Research stage

7. MEMS Mirror LiDAR

7.1 MEMS Technology Overview

MEMS (Micro-Electro-Mechanical Systems) mirrors use microscale mirrors (1-5mm diameter) actuated by electrostatic, electromagnetic, or piezoelectric forces to steer laser beams. They represent the most mature "semi-solid-state" approach.

7.2 MEMS Mirror Types

Type	Actuation	Mirror Size	Scan Angle	Resonant Freq	Durability
Electrostatic comb-drive	Voltage (30-200V)	1-2mm	±10-15°	1-10 kHz	High (no contact)
Electromagnetic	Current coil	2-5mm	±15-25°	0.5-5 kHz	Moderate
Piezoelectric (PZT)	Voltage (1-20V)	1-3mm	±10-20°	1-20 kHz	Good

7.3 MEMS Reliability for Airside

MEMS mirrors are sensitive to:

Mechanical shock: Airport operations involve loading/unloading impacts (5-50g)
- MEMS can handle 1,000-2,000g shock (inherent advantage of microscale)
- Much better than macroscale spinning bearings
Vibration: MEMS resonant frequency (1-20 kHz) well above vehicle vibration spectrum (<500 Hz)
- Vibration isolation requirements minimal
Fatigue: MEMS flexures can achieve 10¹¹-10¹² cycles before failure
- At 5 kHz scan rate, 24/7 operation: ~1.6×10¹¹ cycles/year → 1-10 year fatigue life
- This is the primary MEMS lifetime limiter for 24/7 airside operations

MEMS verdict for airside: Significantly better than mechanical spinning (5-10x lifetime improvement) but still has moving parts subject to fatigue. A bridge technology toward fully solid-state OPA/FPA.

8. Flash LiDAR

8.1 Operating Principle

Flash LiDAR illuminates the entire FoV simultaneously with a single laser pulse or CW modulation, then images the reflected light onto a 2D detector array (like a camera but measuring distance per pixel).

Flash LiDAR:

  ┌─────────┐                        ┌──────────────┐
  │ Laser   │──→ Diffuser ──→ Wide   │              │
  │ Source   │        illumination    │   SCENE      │
  └─────────┘                        │              │
                                     └──────┬───────┘
                                            │
                                     Reflected light
                                            │
                                     ┌──────↓───────┐
                                     │ Imaging Lens  │
                                     └──────┬───────┘
                                            │
                                     ┌──────↓───────┐
                                     │ SPAD Array    │
                                     │ 128×128 to    │
                                     │ 1024×1024     │
                                     │ pixels        │
                                     │               │
                                     │ Each pixel:   │
                                     │ - ToF range   │
                                     │ - intensity   │
                                     └───────────────┘

8.2 Flash LiDAR for Airside Applications

Advantages:

No moving parts, no beam steering electronics → highest reliability
Entire frame captured simultaneously → no motion distortion
Simple optics → lowest cost at volume
Ideal for short-range docking applications (wide FoV, dense point cloud at close range)

Limitations:

Range limited by eye-safe power spread across FoV: typically 20-50m for Class 1
Resolution limited by detector array pixel count (currently 128×128 to 640×640)
Lower SNR per pixel than scanned approaches

Best airside application: Docking sensors. A flash LiDAR at each docking interface provides:

Dense point cloud at 0-10m range
No mechanical wear
Fast frame rate (30-100 Hz)
Cost-effective ($200-500 per unit)

8.3 SPAD Array Detectors

Single-Photon Avalanche Diode (SPAD) arrays are the enabling detector technology for flash LiDAR:

Parameter	2022	2025	Target
Array size	256×256	512×512	1024×1024
Pixel pitch	10-15 μm	6-10 μm	3-5 μm
PDE (Photon Detection Efficiency)	20-30%	30-45%	>50%
DCR (Dark Count Rate)	100-500 cps	50-200 cps	<50 cps
Timing jitter	100-200 ps	50-100 ps	<30 ps
Time gating	Yes	Yes, faster	Sub-ns programmable

9. Airside-Specific Requirements

9.1 Environmental Challenges Unique to Airports

Challenge	Description	Impact on LiDAR	Solid-State Advantage
De-icing spray	Propylene glycol / potassium formate aerosol	Blocks optical path, coats optics	Sealed flat window easier to clean; self-heating with SiPh power dissipation
Jet blast	100+ km/h exhaust at >200°C	Mechanical shock, dust/debris impact	No moving parts to damage; FMCW detects blast velocity
Fuel vapors	JP-A kerosene fumes	Potential optical absorption at some wavelengths	1550nm not affected by JP-A
Tarmac heat shimmer	Radiative heat creates optical turbulence	Beam wander, range noise	FMCW coherent detection less sensitive to turbulence
FOD	Small objects 1-10cm	Requires high angular resolution	OPA adaptive resolution can focus on suspicious areas
Aircraft reflections	Large specular surfaces	Multi-path, saturation	FMCW rejects multi-path via coherence gating
24/7 operations	16-20 hrs/day per vehicle	Wear-out of moving parts	Semiconductor-level reliability (100K+ hrs)
Solar glare	Direct sunlight, tarmac reflection	Saturation, reduced SNR	FMCW coherent detection rejects solar background

9.2 Range and Resolution Requirements

Operational Scenario	Required Range	Required Angular Res	Required Range Precision	Priority
Personnel detection	100m	<0.15°	<5cm	Critical
Aircraft detection	200m	<0.2°	<10cm	Critical
FOD detection	50m	<0.05°	<2cm	High
Precision docking	5m	<0.02°	<5mm	High
GSE detection	150m	<0.2°	<10cm	Critical
Taxiway boundaries	100m	<0.2°	<5cm	Medium
Jet blast detection	80m	N/A (velocity field)	N/A	High (FMCW only)

9.3 Sensor Configuration for Solid-State Migration

CURRENT (Mechanical):                    FUTURE (Solid-State):
                                         
4-8× RoboSense RSHELIOS/RSBP           4× FMCW OPA (long-range 360° coverage)
  - 360° spinning each                  + 4× Flash LiDAR (short-range docking)
  - Overlapping coverage                + 2× FMCW narrow-FoV (forward, rear)
  - No velocity                         
                                         Total: 10 sensors
Total: 4-8 sensors                       + Per-point velocity
                                         + Adaptive resolution
                                         + No moving parts
                                         + Lower total power
                                         
Power: 60-240W                          Power: 30-80W (est.)
Weight: 2.4-4.8 kg                      Weight: 0.3-1.5 kg
Cost: $6K-24K                           Cost: $3K-8K (at volume)
MTBF: ~20K hrs                          MTBF: ~80-100K+ hrs

9.4 Eye Safety at 1550nm

Most solid-state FMCW LiDARs operate at 1550nm (vs. 905nm for most ToF):

Parameter	905nm (ToF)	1550nm (FMCW)
Eye safety limit (MPE)	1 mW/cm²	100 mW/cm²
Retinal hazard	High (focused by cornea onto retina)	Low (absorbed by vitreous humor before retina)
Safe power (Class 1)	~1 mW peak	~10 mW CW
Range at safe power	100-200m (pulsed gain)	100-300m (coherent gain compensates)
Airport relevance	Workers near sensors	100x more margin for ground crew safety

1550nm is strongly preferred for airside — ground crew work within 1-2m of vehicle-mounted sensors. The 100x higher eye-safety limit at 1550nm provides critical safety margin.

10. Reliability and Lifetime Analysis

10.1 Failure Mode Comparison

Component	Mechanical LiDAR	MEMS LiDAR	OPA Solid-State
Bearing failure	Primary failure mode (wear-out)	N/A	N/A
Mirror fatigue	N/A	Eventual (10¹¹ cycles)	N/A
Laser degradation	GaAs 905nm (20K-40K hrs)	Same	InP/InGaAs 1550nm (50K-100K hrs)
Connector fatigue	Rotating joint (weak point)	Fixed	Fixed
Thermal cycling	Differential expansion	Low mass = fast equalization	Monolithic = no differential
Contamination	Seal failure → internal	Sealed package	Hermetic package
Electronics	Driver board	Driver ASIC	Integrated on die

10.2 Projected MTBF

Using MIL-HDBK-217F reliability prediction methodology and adjusting for airside environmental factors (K_env = 2.5 for ground mobile, airfield):

Technology	Base MTBF (benign)	Airside MTBF (K=2.5)	Annual Failure Rate (24/7)
Mechanical spinning	50,000 hrs	20,000 hrs	~35%
MEMS mirror	80,000 hrs	32,000 hrs	~22%
OPA (silicon photonics)	200,000 hrs	80,000 hrs	~9%
Flash (SPAD array)	250,000 hrs	100,000 hrs	~7%
Photonic FPA	250,000+ hrs	100,000+ hrs	~7%

10.3 Fleet-Level Impact

For a 50-vehicle fleet with 6 LiDARs each (300 total sensors):

Technology	Annual Sensor Failures	Replacement Cost/Year	Downtime Hours/Year
Mechanical (current)	~105	$210-525K	~525-1,050 hrs
MEMS	~66	$66-264K	~330-660 hrs
OPA solid-state	~27	$14-54K	~135-270 hrs
Flash	~21	$4-11K	~105-210 hrs

Solid-state migration saves $150-450K/year in sensor replacement for a 50-vehicle fleet, plus 400-800 hours of reduced downtime.

11. Integration with reference airside AV stack Stack

11.1 Point Cloud Format Compatibility

FMCW LiDAR produces extended point clouds with velocity. ROS integration requires extending PointCloud2 message:

python

#!/usr/bin/env python3
"""
ROS driver shim for FMCW solid-state LiDAR.
Extends standard PointCloud2 with per-point radial velocity.
"""

import rospy
import numpy as np
from sensor_msgs.msg import PointCloud2, PointField
from sensor_msgs import point_cloud2
from std_msgs.msg import Header


# Extended point fields for FMCW LiDAR
FMCW_POINT_FIELDS = [
    PointField('x', 0, PointField.FLOAT32, 1),
    PointField('y', 4, PointField.FLOAT32, 1),
    PointField('z', 8, PointField.FLOAT32, 1),
    PointField('intensity', 12, PointField.FLOAT32, 1),
    PointField('radial_velocity', 16, PointField.FLOAT32, 1),  # FMCW-specific
    PointField('snr', 20, PointField.FLOAT32, 1),  # Signal-to-noise ratio
    PointField('ring', 24, PointField.UINT16, 1),  # Scan line index
    PointField('time', 26, PointField.FLOAT32, 1),  # Per-point timestamp
]


class FMCWLidarBridge:
    """Bridge FMCW solid-state LiDAR to ROS PointCloud2."""
    
    def __init__(self):
        self.pub = rospy.Publisher('/fmcw_lidar/points', PointCloud2, queue_size=2)
        self.pub_velocity = rospy.Publisher('/fmcw_lidar/velocity_cloud', 
                                            PointCloud2, queue_size=2)
        
        # Backward compatibility: publish standard XYZI cloud on legacy topic
        self.pub_compat = rospy.Publisher('/rslidar_points', PointCloud2, queue_size=2)
    
    def publish_scan(self, fmcw_data):
        """
        Convert FMCW LiDAR native data to ROS PointCloud2.
        
        Publishes on two topics:
        1. Full FMCW cloud with velocity (for new FMCW-aware pipeline)
        2. Standard XYZI cloud (backward compatible with existing PointPillars/GTSAM)
        """
        header = Header()
        header.stamp = rospy.Time.now()
        header.frame_id = 'fmcw_lidar'
        
        # Full FMCW point cloud
        full_cloud = point_cloud2.create_cloud(header, FMCW_POINT_FIELDS, fmcw_data)
        self.pub.publish(full_cloud)
        
        # Backward-compatible XYZI cloud (drop velocity, SNR)
        xyzi_data = fmcw_data[:, :4]  # x, y, z, intensity only
        compat_fields = FMCW_POINT_FIELDS[:4]
        compat_cloud = point_cloud2.create_cloud(header, compat_fields, xyzi_data)
        self.pub_compat.publish(compat_cloud)

11.2 Velocity-Enhanced Perception Pipeline

CURRENT PIPELINE:                    FMCW-ENHANCED PIPELINE:
                                     
PointCloud (XYZI)                    PointCloud (XYZI + V)
      │                                    │
      ↓                                    ├──→ Velocity segmentation
  PointPillars                             │    (static vs dynamic in
  (Detection)                              │     single scan, no tracker)
      │                                    ↓
      ↓                              PointPillars+V
  CenterPoint                        (Velocity-augmented detection)
  Tracker (MOT)                            │
      │                                    ↓
      ↓                              Simplified tracker
  Velocity estimate                  (velocity already measured)
  (3-5 frame delay)                        │
      │                                    ↓
      ↓                              Instant velocity
  Frenet planner                     (zero-frame delay!)
                                           │
                                           ↓
                                     Frenet planner
                                     (velocity-aware costs)

11.3 GTSAM Localization Enhancement

FMCW velocity data improves ego-velocity estimation for GTSAM:

cpp

// Add FMCW ego-velocity factor to GTSAM factor graph
// Uses per-point radial velocity to estimate vehicle velocity

#include <gtsam/navigation/NavState.h>
#include <gtsam/nonlinear/NonlinearFactor.h>

class FMCWVelocityFactor : public gtsam::NoiseModelFactor1<gtsam::NavState> {
    gtsam::Vector3 measured_velocity_;  // From FMCW Doppler
    
public:
    FMCWVelocityFactor(gtsam::Key key, const gtsam::Vector3& measured_v,
                       const gtsam::SharedNoiseModel& model)
        : NoiseModelFactor1(model, key), measured_velocity_(measured_v) {}
    
    gtsam::Vector evaluateError(const gtsam::NavState& state,
                                 boost::optional<gtsam::Matrix&> H) const override {
        // Error = predicted velocity - measured velocity
        gtsam::Vector3 predicted_v = state.velocity();
        if (H) {
            // Jacobian: d(error)/d(state) for velocity component
            *H = gtsam::Matrix::Zero(3, 9);
            (*H).block<3,3>(0, 6) = gtsam::Matrix3::Identity();  // dv/dv
        }
        return predicted_v - measured_velocity_;
    }
};

// Usage in GTSAM factor graph:
// 1. Estimate ego-velocity from static-point FMCW velocities
// 2. Static points: radial_velocity ≈ -v_ego · point_direction
// 3. RANSAC to separate static/dynamic, estimate v_ego
// 4. Add FMCWVelocityFactor with estimated v_ego

11.4 Adaptive Resolution for Active Perception

OPA beam steering enables dynamic resolution allocation:

python

class AdaptiveResolutionManager:
    """
    Manage OPA LiDAR scan pattern based on scene context.
    
    Unlike mechanical LiDAR (fixed scan pattern), OPA can allocate
    more points to regions of interest and fewer to empty space.
    """
    
    # Region-of-interest priorities for airside
    ROI_PRIORITIES = {
        'aircraft_zone': 1.0,       # Highest: near aircraft
        'personnel_area': 0.9,      # Very high: where people work  
        'active_taxiway': 0.7,      # High: traffic areas
        'docking_approach': 1.0,    # Highest: precision needed
        'empty_taxiway': 0.2,       # Low: open space
        'sky': 0.0,                 # None: waste of points
    }
    
    def compute_scan_pattern(self, scene_context, total_points_budget=100000):
        """
        Allocate point budget across FoV based on scene context.
        
        Returns: List of (azimuth, elevation, dwell_time) per point
        """
        regions = scene_context.get_regions()
        
        # Allocate points proportional to priority × solid angle
        total_weighted_area = sum(
            r.solid_angle * self.ROI_PRIORITIES.get(r.type, 0.3)
            for r in regions
        )
        
        scan_points = []
        for region in regions:
            priority = self.ROI_PRIORITIES.get(region.type, 0.3)
            n_points = int(total_points_budget * 
                          (region.solid_angle * priority) / total_weighted_area)
            
            # Generate uniform point grid within region
            points = region.generate_uniform_grid(n_points)
            scan_points.extend(points)
        
        return scan_points

12. Cost Roadmap and Market Dynamics

12.1 LiDAR Cost Trajectory

Year	Mechanical Spinning	MEMS	OPA Solid-State	Flash
2020	$4,000-10,000	$1,000-3,000	N/A (lab only)	$2,000-5,000
2023	$1,500-5,000	$500-1,500	$5,000+ (prototype)	$1,000-3,000
2025	$1,000-3,000	$300-800	$1,000-3,000	$500-1,500
2027 (proj.)	$800-2,000	$200-500	$200-800	$200-500
2030 (proj.)	$500-1,500	$100-300	$50-200	$100-300

12.2 Volume Dependency

Cost per unit ($)
    │
10K ┤ ○ 
    │  ╲
 5K ┤   ╲  Mechanical (limited scaling)
    │    ╲─────────────────────────────
 2K ┤     ╲
    │      ○ 
 1K ┤       ╲  MEMS (moderate scaling)
    │        ╲──────────────────────
500 ┤         ╲
    │          ╲
200 ┤           ○  OPA/SiPh (semiconductor scaling)
    │            ╲
100 ┤             ╲─────────────────
 50 ┤              ╲
    │               ╲_______________
    └──┬──┬──┬──┬──┬──┬──┬──┬──┬──→
      1K 10K 100K 1M 10M         Units/year
      
Key insight: Silicon photonics follows semiconductor cost curves.
At automotive volumes (1M+ units/year), OPA LiDAR approaches $50-100.
At reference airside AV stack volumes (100-1000 units/year), OPA premium still significant.
Strategy: Use automotive-volume sensors designed for cars, adapt for airside.

12.3 Market Consolidation (2025-2028)

The LiDAR market is consolidating rapidly:

Ibeo: Filed insolvency (2022), acquired by MicroVision
Velodyne + Ouster: Merged (2023), now focused on industrial/robotics
Quanergy: Bankrupt (2022)
Cepton: Acquired by Koito (2023)
Innoviz: Secured BMW production deal, largest MEMS deployment
Aeva: Only pure-play FMCW company with automotive design win
Voyant: Newest entrant, most aggressive silicon photonics approach

Implication for reference airside AV stack: Don't bet on a single vendor. Design sensor interfaces that can swap between RoboSense (current), Aeva (FMCW mid-term), and silicon photonics OPA (long-term). The ROS PointCloud2 abstraction already provides this.

13. Migration Strategy: Mechanical to Solid-State

13.1 Phased Migration

Phase 1 (2025-2026): EVALUATE
├── Add 1-2 Aeva Atlas (FMCW) alongside existing RoboSense
├── Run dual-stack: existing pipeline + FMCW-enhanced pipeline
├── Validate velocity data utility for tracking, jet blast detection
├── No changes to safety-critical pipeline
└── Cost: $5-10K hardware + $15-20K integration

Phase 2 (2026-2027): AUGMENT
├── Replace 2 of 6 RoboSense with FMCW (forward-facing)
├── Add flash LiDAR for docking (2-4 per vehicle)
├── Modify PointPillars to use velocity channel
├── FMCW velocity feeds GTSAM ego-velocity factor
└── Cost: $10-20K hardware + $25-35K software

Phase 3 (2027-2028): PRIMARY TRANSITION
├── Majority FMCW sensors (4 of 6 positions)
├── Keep 2 mechanical as redundancy during transition
├── Full velocity-aware perception pipeline
├── Adaptive resolution active perception
└── Cost: $15-25K hardware + $20-30K software

Phase 4 (2028-2030): FULL SOLID-STATE
├── All sensors solid-state (OPA + flash)
├── Remove mechanical LiDAR entirely
├── Per-point velocity as primary input to planner
├── Fleet-wide sensor standardization
└── Cost: $10-20K per vehicle (sensor swap)

13.2 Backward Compatibility Requirement

Critical: Each phase must maintain backward compatibility with existing perception pipeline. The ROS PointCloud2 message format enables this — FMCW sensors publish standard XYZI on the same topic, with velocity as an optional additional field.

python

class DualStackPerception:
    """
    Run existing ToF pipeline and FMCW-enhanced pipeline in parallel.
    Simplex pattern: FMCW-enhanced as AC, existing ToF as BC.
    """
    
    def __init__(self):
        # Existing pipeline (backward compatible)
        self.tof_pipeline = PointPillarsDetector()  # Standard XYZI input
        
        # New FMCW-enhanced pipeline
        self.fmcw_pipeline = VelocityAwareDetector()  # XYZIV input
        
        # Safety monitor compares both
        self.safety_monitor = PerceptionSafetyMonitor()
    
    def process(self, cloud_msg):
        """
        Dual-stack processing during migration phases.
        """
        # Extract standard XYZI for backward-compatible pipeline
        xyzi = extract_xyzi(cloud_msg)
        tof_detections = self.tof_pipeline.detect(xyzi)
        
        # Extract XYZIV for FMCW-enhanced pipeline (if velocity available)
        if has_velocity_field(cloud_msg):
            xyziv = extract_xyziv(cloud_msg)
            fmcw_detections = self.fmcw_pipeline.detect(xyziv)
            
            # Safety monitor: cross-check both pipelines
            if self.safety_monitor.agree(tof_detections, fmcw_detections):
                return fmcw_detections  # Use enhanced results
            else:
                rospy.logwarn("Perception disagreement — using conservative ToF pipeline")
                return tof_detections  # Fallback to proven pipeline
        else:
            return tof_detections  # No FMCW data available

14. Implementation Roadmap

14.1 Timeline and Costs

Phase	Duration	Hardware Cost	Software Cost	Total	Key Deliverable
Phase 1: Evaluation	8 weeks	$5-10K	$15-20K	$20-30K	FMCW data characterization report
Phase 2: Augmentation	12 weeks	$10-20K	$25-35K	$35-55K	Velocity-enhanced PointPillars, flash docking
Phase 3: Primary transition	16 weeks	$15-25K	$20-30K	$35-55K	Majority FMCW fleet, adaptive resolution
Phase 4: Full solid-state	12 weeks	$10-20K/vehicle	$10-15K	$20-35K + per-vehicle	All-solid-state fleet
Total	48 weeks	$40-75K + fleet	$70-100K	$110-175K + fleet

14.2 Risk Mitigation

Risk	Probability	Impact	Mitigation
FMCW range insufficient for airside	Low	High	Evaluate multiple vendors; 200m is sufficient, most claim 300m+
OPA resolution inadequate	Medium	Medium	MEMS as bridge; flash LiDAR for close-range
Software integration complexity	Medium	Medium	Dual-stack approach allows gradual migration
Vendor bankruptcy/exit	Medium	High	Vendor-agnostic ROS interface; multi-source strategy
Point rate too low for PointPillars	Low	High	Accumulate 2-3 scans; OPA point rates increasing rapidly
1550nm atmospheric absorption	Low	Low	1550nm absorption is minimal in normal atmosphere; FMCW coherent gain compensates

15. Key Takeaways

Solid-state LiDAR saves $150-450K/year for a 50-vehicle fleet through reduced sensor replacement, less downtime, and lower maintenance labor
FMCW provides per-point velocity at zero additional latency — eliminates 300-500ms tracking delay for velocity estimation, transformative for jet blast detection and approaching-object early warning
1550nm eye safety is critical for airside — 100x higher safe power limit vs 905nm protects ground crew working within 1-2m of sensors
Voyant Helium (photonic focal plane array) is the long-term target — truly no moving parts, semiconductor scaling economics, <150g per sensor. Prototype at CES 2026
Aeva Atlas is the near-term practical choice — FMCW + MEMS, 300m range, automotive-grade, production 2025. Best bridge technology while OPA matures
Migration is incremental, not big-bang — dual-stack Simplex pattern (FMCW-enhanced AC, existing ToF BC) allows gradual transition with zero safety regression
Adaptive OPA resolution enables foveated perception — dynamically allocate more points to aircraft/personnel zones, fewer to empty taxiway. 2-5x effective resolution improvement in regions that matter
Flash LiDAR solves docking — wide-FoV, dense, no moving parts, $200-500/unit. Complementary to long-range FMCW for docking applications requiring ±5cm precision
Silicon photonics cost curves follow semiconductors — current $1,000-3,000 per FMCW sensor will reach $50-200 at automotive volumes (2028-2030). reference airside AV stack volumes (100-1000/year) benefit from automotive-scale pricing
FMCW velocity data improves GTSAM localization — direct ego-velocity measurement from static-point Doppler provides additional factor for fusion alongside IMU, wheel odometry, and RTK-GPS

16. References

Papers and Standards

Poulton, C.V. et al., "Large-scale silicon nitride nanophotonic phased arrays at infrared and visible wavelengths," Optics Letters, 2017
Rogers, C. et al., "A universal 3D imaging sensor on a silicon photonics platform," Nature, 2021
Hsu, C.P. et al., "A Review and Perspective on Optical Phased Array for Automotive LiDAR," IEEE Journal of Selected Topics in Quantum Electronics, 2021
Li, B. et al., "A large-scale microelectromechanical-systems-based silicon photonics LiDAR," Nature, 2022
Sun, X. et al., "Si Photonics FMCW LiDAR Chip with Solid-State Beam Steering by Interleaved Coaxial Optical Phased Array," Micromachines, 2023
Isaac, B.J. et al., "Photonic-electronic integrated circuit-based coherent LiDAR engine," Nature Communications, 2024
Fan, G. et al., "Progress and prospects for LiDAR-oriented optical phased arrays based on photonic integrated circuits," npj Nanophotonics, 2025
Luo, G. et al., "Requirements for next-generation integrated photonic FMCW LiDAR sources," Nature Communications, 2025

Products and Companies

Aeva, "Atlas — Automotive-Grade 4D LiDAR," Product Specification, 2025
Aeva, "Aeries II — Camera-Level Resolution 4D LiDAR," Product Specification, 2026
Voyant Photonics, "Helium Platform — Fully Solid-State 4D FMCW LiDAR," Press Release, December 2025
SiLC Technologies, "Eyeonic Vision System — FMCW LiDAR with Polarization," Product Specification, 2024
Innoviz Technologies, "InnovizTwo — Automotive-Grade MEMS LiDAR," Product Specification, 2024

Market Analysis

Yole Group, "LiDAR for Automotive and Industrial 2025," Market Report, 2025
PatSnap, "LiDAR Sensor Technology Landscape for Autonomous 2026," Analysis, 2026

20-av-platform/sensors/sensor-degradation-health-monitoring.md — Sensor health monitoring framework
20-av-platform/sensors/multi-lidar-calibration.md — Multi-LiDAR calibration (applies to mixed ToF+FMCW fleet)
30-autonomy-stack/perception/overview/active-perception-sensor-scheduling.md — Adaptive resolution scheduling (OPA enables this)
30-autonomy-stack/planning/autonomous-docking-precision-positioning.md — Docking use case for flash LiDAR
30-autonomy-stack/perception/overview/radar-lidar-fusion-adverse-weather.md — Weather resilience (FMCW LiDAR + 4D radar complementary)

SLAM Methods

Methods

Solid-State LiDAR and Photonic Integrated Circuits for Airside Autonomous GSE ​

Table of Contents ​

1. Introduction & Motivation ​

1.1 The Mechanical LiDAR Problem for 24/7 Airside Operations ​

1.2 The Solid-State Promise ​

1.3 Technology Readiness ​

2. LiDAR Beam Steering Technologies ​

2.1 Technology Comparison ​

2.2 Beam Steering Methods Compared ​

2.3 Scan Patterns ​

3. FMCW vs ToF: Measurement Principles ​

3.1 Time-of-Flight (Current reference airside AV stack Approach) ​

3.2 Frequency-Modulated Continuous Wave (FMCW) ​

3.3 Velocity Measurement Value ​

4. Silicon Photonics Integration ​

4.1 What is Silicon Photonics? ​

4.2 Key Components on Silicon ​

4.3 Material Platforms ​

4.4 Integration Levels ​

5. Solid-State LiDAR Products and Specifications ​

5.1 FMCW Solid-State Products (2024-2026) ​

5.2 Semi-Solid-State Products (In Production) ​

5.3 Comparison with Current reference airside AV stack Sensors ​

6. Optical Phased Arrays (OPA) ​

6.1 Operating Principle ​

6.2 OPA Performance State-of-the-Art (2025-2026) ​

6.3 2D Steering Approaches ​

6.4 OPA Challenges and Solutions ​

7. MEMS Mirror LiDAR ​

Solid-State LiDAR and Photonic Integrated Circuits for Airside Autonomous GSE

Table of Contents

1. Introduction & Motivation

1.1 The Mechanical LiDAR Problem for 24/7 Airside Operations

1.2 The Solid-State Promise

1.3 Technology Readiness

2. LiDAR Beam Steering Technologies

2.1 Technology Comparison

2.2 Beam Steering Methods Compared

2.3 Scan Patterns

3. FMCW vs ToF: Measurement Principles

3.1 Time-of-Flight (Current reference airside AV stack Approach)

3.2 Frequency-Modulated Continuous Wave (FMCW)

3.3 Velocity Measurement Value

4. Silicon Photonics Integration

4.1 What is Silicon Photonics?

4.2 Key Components on Silicon

4.3 Material Platforms

4.4 Integration Levels

5. Solid-State LiDAR Products and Specifications

5.1 FMCW Solid-State Products (2024-2026)

5.2 Semi-Solid-State Products (In Production)

5.3 Comparison with Current reference airside AV stack Sensors

6. Optical Phased Arrays (OPA)

6.1 Operating Principle

6.2 OPA Performance State-of-the-Art (2025-2026)

6.3 2D Steering Approaches

6.4 OPA Challenges and Solutions

7. MEMS Mirror LiDAR