LiDAR technology has become one of the core tools in the field of autonomous driving in recent years. As a sensor that measures distance and generates high-precision 3D maps by emitting laser beams and receiving reflected signals, LiDAR offers unparalleled environmental perception capabilities. This article will introduce the development of LiDAR applications in autonomous driving and the related solutions launched by onsemi.
LiDAR technology matures to pave the way for full autonomous driving
Compared to traditional cameras and radar, LiDAR can accurately capture details of the surrounding environment under various lighting and weather conditions. This feature makes it a key element in achieving true autonomous driving functionality. By precisely measuring distances and identifying objects, LiDAR not only enhances the vehicle’s navigation and obstacle avoidance capabilities but also improves overall driving safety and reliability.
The development trajectory of LiDAR in autonomous driving applications has evolved from early exploration to widespread adoption. Initially, LiDAR systems were primarily used for scientific research and military purposes, such as terrain mapping and target detection. As technology progressed, LiDAR was gradually introduced into the autonomous driving field. At first, LiDAR was mainly used for concept validation and testing in experimental environments.
The first-generation LiDAR systems consisted mainly of high-cost and bulky equipment, typically mounted on the roof of vehicles. These systems covered a 360-degree field of view through rotating mirrors or rotating lasers, generating point cloud data of the vehicle's surroundings. However, these early LiDAR systems were difficult to commercialize on a large scale due to their high costs, large size, and high power consumption.
With technological advancements, the cost of LiDAR sensors has gradually decreased, their size has become smaller, and their performance has continuously improved. The emergence of solid-state LiDAR was a significant milestone. This technology reduced the use of mechanical components, lowering system complexity and costs while increasing reliability and durability. New technologies such as Microelectromechanical Systems (MEMS) LiDAR and Optical Phased Array (OPA) LiDAR have also begun to appear, further promoting the application of LiDAR in autonomous driving.
As the technology matured, major automakers and tech companies started large-scale road testing and validation efforts. Companies like Waymo, Uber, Tesla, and others invested heavily in testing the performance and reliability of LiDAR systems under real-world road conditions. These tests not only validated LiDAR’s sensing capabilities in complex traffic environments but also promoted the advancement and optimization of autonomous driving algorithms.
In recent years, LiDAR technology has gradually moved towards commercial application, with an increasing number of autonomous vehicles and advanced driver assistance systems (ADAS) starting to incorporate LiDAR sensors. Leading LiDAR manufacturers such as Velodyne, Luminar, Innoviz, and others have launched multiple LiDAR products suitable for mass production vehicles. These products play a crucial role in enhancing safety, improving vehicle environmental perception capabilities, and achieving highly automated driving.
As LiDAR technology continues to advance and costs further decrease, LiDAR is expected to play an increasingly important role in autonomous vehicles. Future LiDAR systems will be more compact, cost-effective, and high-performance, working in conjunction with other sensors (such as cameras and radar) to achieve more comprehensive and reliable environmental perception. This will pave the way for the realization of full autonomous driving, driving profound changes in the automotive industry.
LiDAR sensor products using SiPM technology
Recognizing the demand in the LiDAR application market, onsemi has partnered with leading automotive LiDAR manufacturer Innoviz to launch InnovizOne, the first commercial LiDAR product from Innoviz aimed at the automotive market. InnovizOne is a product that utilizes LiDAR sensing technology and can be combined with other sensor modalities such as cameras and radars to enable true hands-free, eyes-off autonomous driving in passenger cars.
InnovizOne incorporates onsemi's ArrayRDM-0112, a 12-channel silicon photomultiplier (SiPM) array, to detect returning laser pulses and measure distances to the surrounding environment, marking a significant advancement in the industry. SiPM technology, rooted in high-volume CMOS silicon foundry processes, has clear advantages over the higher-cost detectors used in 1550 nm short-wave infrared LiDAR. SiPMs utilize very high gain to detect single photons, making them superior to the low-sensitivity avalanche photodiode detectors that limit LiDAR range measurement.
Innoviz is primarily focused on providing high-performance LiDAR at the lowest possible cost, enabling broader adoption by automotive OEMs. This is achieved through their use of a 905 nm-based near-infrared direct time-of-flight (dToF) LiDAR architecture.
In 2018, BMW and Magna jointly selected InnovizOne as the LiDAR technology provider for their future passenger vehicle platforms. After several years of development, InnovizOne entered mass production in 2023. BMW equipped its flagship 7-Series vehicles (such as the i7) with LiDAR to enable BMW Personal Pilot, offering true Level 3 "eyes-off" autonomous driving, which launched in Germany in the spring of 2024. Innoviz recently announced that BMW will also extend InnovizOne to its mainstream 5-Series vehicles in the Chinese market. This exciting development signifies that LiDAR is becoming suitable not only for high-end luxury vehicles but also for mainstream models, providing safe autonomous driving sensors to the mass market.
Rugged and durable automotive-grade solid-state LiDAR sensor
The InnovizOne is a solid-state LiDAR sensor solution designed specifically for automotive manufacturers and companies involved in robot taxis, shuttles, and delivery services, aiming for scalable production and reliable operation in autonomous driving applications. This automotive-grade sensor is engineered to be rugged, affordable, reliable, low-power, lightweight, and high-performance, ensuring the safety of passengers and pedestrians when seamlessly integrated into Level 3-5 autonomous vehicles.
InnovizOne can adapt to sunlight and weather conditions, providing rich 3D point clouds at distances of up to 250 meters. When combined with Innoviz's perception software, the sensor achieves superior object detection, classification, and tracking with unmatched angular resolution.
The InnovizOne seamlessly integrates with perception software, designed and developed to meet the highest standards of the automotive industry, InnovizOne has achieved the necessary certifications and approvals, including ISO 9001:2015 certification and early implementation of ISO 26262 (functional safety) in its components (ASIC, detectors, and MEMS), as well as the LiDAR hardware and software systems. Innoviz has selected manufacturers throughout its supply chain to meet IATF 16949 requirements and follows ASPICE standards for software development.
InnovizOne supports an angular resolution of 0.1°x0.1° (HxV), with a pre-configured frame rate of 10 or 15 FPS, and a detection range from 1m to 250m. It offers a maximum field of view (FoV) of 115°x25° (HxV) and measures 45x111.4x97.9 mm (HxWxD), conforming to the ISO 26262 ASIL B(D) standard.
The InnovizOne features a pre-configured Region of Interest (ROI) at the center of the LiDAR's field of view, where laser energy is concentrated to extend the detection range. The ROI measures 20°x8° (HxV), without affecting the data interface's bandwidth, resolution, or frame rate. InnovizOne can return multiple reflections per pixel, recording and storing multiple points in the 3D environment—critical when laser pulses hit raindrops, snowflakes, or multiple objects along the path.
Due to its continuous pixel feature, InnovizOne's LiDAR scanning pattern has no gaps, which is crucial for building a safe autonomous vehicle perception system. Without this, the system might miss small objects on the road surface related to collisions or humans located in gaps within the 3D point cloud data, potentially leading to catastrophic consequences.
InnovizOne provides a uniformly distributed vertical FoV, maintaining consistent resolution across the entire vertical field. This capability allows the system to capture more data compared to sensors that focus only on the center (horizon) and lose data towards the edges. InnovizOne’s vertical field also has a panning feature to support mounting tolerances and different driving conditions such as vehicle loading. It can handle various environmental lighting conditions, such as direct sunlight, blinding lights from oncoming vehicles, and adverse weather conditions like rain, ensuring it adapts to all ambient light sources. InnovizOne is applicable in a wide range of fields, including consumer vehicles, robot taxis, shuttles, delivery services, trucking, logistics, sidewalk delivery, industrial drones, and heavy machinery.
SiPM for automotive NIR-enhanced LiDAR applications
onsemi has introduced the ArrayRDM-0112A20-QFN, a silicon photomultiplier (SiPM) specifically designed for automotive NIR-enhanced LiDAR applications. The SiPM is a high-gain, single-photon-sensitive sensor capable of detecting light from visible to near-infrared wavelengths. The ArrayRDM-0112A20-QFN is a monolithic 1 × 12 SiPM pixel array based on the market-leading RDM process, specifically developed to achieve high photon detection efficiency (PDE) at 905/940 nm NIR wavelengths, commonly used in LiDAR and 3D direct time-of-flight (dToF) ranging applications.
The ArrayRDM-0112A20-QFN array comes in a robust QFN package, providing access to 12 individual pixels. To meet the requirements of automotive LiDAR applications, this product will comply with AEC-Q102 standards. The ArrayRDM-0112A20-QFN features high gain and detection efficiency, meeting automotive standards. It utilizes a 1 × 12 pixel array format with a PDE of 16% at 905 nm, operates at a 30 V bias voltage, and pixel size only 0.47 mm x 1.12 mm, housed in a QFN package (10 mm x 5.2 mm).
To accelerate product development, onsemi has also developed an evaluation board (ArrayRDM-0112A20-GEVB) for this product. The ArrayRDM-0112A20-QFN can be applied in 3D ranging and sensing, automotive LiDAR, industrial LiDAR, consumer 3D imaging, and robotics, with typical end products being scanning LiDAR systems.
Conclusion
The development of LiDAR in autonomous driving showcases the mutual drive of technological advancements and market demands. From early high-cost experimental equipment to today's mass-produced applications, LiDAR is continually breaking through technical bottlenecks, providing robust support for achieving true autonomous driving. onsemi has introduced SiPM technology suitable for automotive NIR (Near Infrared) enhanced LiDAR applications. This technology can lower the application costs of autonomous driving technology, enhance LiDAR performance, and improve integration with other vehicle systems. Thus, it is an ideal choice for developing autonomous driving applications.