Sensor Suite Comparison by Platform
How many of each sensor type each self-driving system uses. Cameras see color, radar sees through fog, lidar (laser-based 3D sensing) builds precise depth maps, ultrasonics handle parking. Tesla uses cameras only; Waymo stacks all four.
PlatformSAECamLidarRadarUltrasonicTotalCompute
Waymo Driver (6th Gen)
Long-range, mid-range, short-range, and perimeter lidar. 360-degree coverage.
Baidu Apollo RT6
Purpose-built robotaxi; production cost ~$37K per unit.
Tesla FSD HW4
Vision-only approach. Removed ultrasonic (2022) and radar (2021), radar restored in HW4.
Mobileye SuperVision
Camera-primary system. Optional lidar in Chauffeur (L3+) config.
NIO NAD (ET7)
Top-mounted 360-degree lidar (Innovusion Falcon). Full sensor fusion.
XPeng XNGP (P7+)
Dual lidar on P7+. Newer models transitioning to vision-only.
Lucid DreamDrive Pro
32-sensor suite; most sensors of any consumer vehicle.
AD Software Platforms & OEM Adoption
Off-the-shelf self-driving software platforms (NVIDIA DriveWorks, Mobileye SuperVision, Qualcomm Ride) and which automakers (OEMs — Original Equipment Manufacturers like Ford, BMW, Geely) have licensed each one. Shows the build-vs-buy split across the industry.
NVIDIA DRIVE(Orin, Thor)
9 OEMsOEMSystem / IntegrationSince
Mercedes-BenzDrive Pilot (L3)2022
JLRADAS Stack2023
BYDDiPilot2023
Hyundai / KiaHighway Driving Pilot2024
LucidDreamDrive Pro2022
XPengXNGP2021
NIONAD2022
Li AutoNOA2023
Volvo / PolestarPilot Assist+2024
Mobileye EyeQ / SuperVision(EyeQ5, EyeQ6H)
7 OEMsOEMSystem / IntegrationSince
BMWHighway Assistant2021
FordBlueCruise (Gen 1)2021
Volkswagen GroupIQ.Drive2022
Zeekr / GeelySuperVision2023
SAICPilot Assist2022
StellantisHighway Assist2023
NissanProPilot Assist2020
Qualcomm Snapdragon Ride(SA8295P, SA8775P)
4 OEMsOEMSystem / IntegrationSince
GM / CadillacUltra Cruise2023
BMWNew Class ADAS (2025+)2025
HondaSENSING 360+2026
StellantisSTLA SmartCockpit2024
Autonomous Driving Hardware Cost Trends
Per-unit cost trend for the three most expensive sensors in a self-driving car: lidar (laser-based 3D sensor), the onboard AI compute chip, and cameras. Lidar dropped from $75K+ in 2017 to under $500 in mass production — the key reason robotaxis are commercially viable today. (Sources: industry estimates, company filings)
Indicative per-unit costs at volume. Sources: industry estimates (Yole Intelligence, analyst reports, company filings).