LiDAR LiDAR systems: costs, integration, and major manufacturers
LiDAR is quickly becoming a "must-have" technology for self-driving vehicles. Until now, a large footprint, high cost, and other limitations have prevented LiDAR from being easily integrated into production autos. But all this is about to change as industry players make deals with each other and rapidly release new products.
Similar to radar, LiDAR is a measurement and rangefinding process originally developed in the 1960s for military and aerospace applications where precise accuracy over long distances was needed. This technology ultimately found its way into meteorology and robotics.
Clearly seen as a valuable tool for self-driving cars, LiDAR's potential integration into vehicles may create a marketplace worth as much as $2.5 billion by 2030.
In a recent article at MES Insights, we provided an overview of how LiDAR works and what its applications are for autonomous vehicles.
This article details price points and specifics of LiDAR integration as well as going over who the major industry players are and a timeline of commercial announcements.
Although LiDAR has come a long way since its beginnings in the 1960s, it is not yet a mature, low-cost, or truly tiny-footprint technology.
It's true that the bulky, roof-mounted LiDAR sensors seen on early self-driving car prototypes such as those from Google or Apple are slowly giving way to low-profile sensors that are mountable on a car's hood, fenders, or front grille.
But the high cost of LiDAR remains a major obstacle to mass integration. "One of the biggest roadblocks to bringing autonomous vehicles to the mainstream has been the lack of a high-performance and low-cost LiDAR that can scale to millions of units per year," says Soroush Salehian, the co-founder of LiDAR sensor maker Aeva. Until recently, LiDAR sensors and systems were priced in the thousands or even tens of thousands of dollars.
However, these price points are beginning to be broken. In 2016, Canadian supplier LeddarTech released an automotive LiDAR sensor that sold for below $500, a significant industry first at the time. In 2020, Velodyne announced it would be introducing a $100 LiDAR sensor applicable for use in cars.
Integration of LiDAR
But sensors are only part of a complete LiDAR system. Hardware and software are still needed to process the data LiDAR sensors receive.
Largely because of its high cost, fleets of so-called "robo-taxis" have been seen as potential first mass-scale adopters of LiDAR, rather than production vehicles meant for consumers. Irish LiDAR vendor Aptiv, for instance, is currently helping to operate a 30-car autonomous taxi fleet in Las Vegas in partnership with ride-sharing firm Lyft.
Waymo claims to have reduced the cost of LiDAR systems for its vehicles by 90 percent by developing its own proprietary hardware. Previously, it had purchased Velodyne LiDAR systems for its cars at a cost of $75,000 per unit.
In 2016, ride-sharing company Uber acquired autonomous truck startup Otto, co-founded by former Waymo employee Anthony Levandowski in 2016. In 2017, Waymo sued Uber, claiming that Levandowski had stolen trade secrets related to autonomous driving and Waymo's proprietary LiDAR.
In 2018, Uber shut down Otto to focus on building self-driving cars within its Advanced Technologies Group (ATG) instead of trucks. In 2019, Waymo began selling its $7,500 LiDar sensor, named the Laser Bear Honeycomb, to other companies, providing that they promised not to compete with Waymo's robo-taxi service.
Beyond cost, however, it's almost a certainty that as it advances, LiDAR will be proven technically superior to radar and video cameras for autonomous driving.
Yet self-driving vehicles of the future (those functioning at SAE Autonomy Levels 3 and above) will likely fuse data from all three systems. Why? Fusion brings two benefits. The first is that a richer, more diverse set of data sources will produce greater navigation accuracy. The second is that multiple systems provide safety redundancy should any one of them have an error or an outage.
Who makes LiDAR?
A number of firms make automotive LiDAR equipment. Companies producing automotive LiDAR sensors include:
- Velodyne (United States)
- Ouster (United States)
- Luminar (United States)
- Baraja (Australia)
- Innoviz (Isreal)
- Valeo (France)
- Sick (Germany)
- Infineon (Germany)
- Benewake (China)
- Livox (China)
- Denso (Japan)
- Hokuyo (Japan)
- Pioneer (Japan)
Companies producing software and other components needed to effect obstacle detection and avoidance include:
- Autonomous Solutions (United States)
- Aptiv (Ireland)
Companies integrating all of the above into complete solutions include:
- Xenomatix (Belgium)
- Ibeo (Germany)
- ContinentalContinental (Germany)
- LeddarTech (Canada)
- Quanergy (United States)
- AEye (United States)
- Robosense (China)
- Hesai (China)
Future players in the growing automotive LIDAR marketplace may include:
- First Sensor (Germany)
- Bosch (Germany)
- Sony (Japan)
- Phantom Intelligence (Canada)
- BorgWarner (formerly Delphi) (United States)
- Intel/Mobileye (United States)
- Lumotive (United States)
- Sense Photonics (United States)
- Garmin (United States)
- Apple (United States)
It's notable that as recently as 2019, Elon Musk, the founder of electric car company Tesla, had shied away from LiDAR, instead of focusing on "2D" camera technology for his company's self-driving cars and prototypes. Many industry observers, including auto research firm Navigant, believe this approach is wrong and may compromise passenger safety.
In mid-2019, researchers from cybersecurity firm McAfee's Advanced Threat Research division used black tape to deface a highway speed sign. They distorted the posted 35 mph speed limit to convince an approaching pre-2017 Tesla Model S vehicle into thinking the speed limit was 85 mph. This caused the Tesla vehicle's Autopilot to attempt to speed up by 50 mph. Although the Model S's Autopilot systems have been upgraded since then, this test showed off marked fallibility of the car's 2D Mobileye electronic vision system. This demonstrates why 3D sensing systems such as LiDAR may be better suited to govern self-driving vehicles' speed.
Like Tesla, carmaker Nissan—still smarting from a financial scandal involving its former chairman Carlos Ghosn—has taken an anti-LiDAR stance, choosing to incorporate sonar in addition to radar and cameras for navigation. "It would be fantastic if LiDAR technology was at the level that we could use it in our systems, but it's not. There's an imbalance between its cost and its capabilities," said Tetsuya Iijima of Nissan's automated driving technology group.
What vehicle manufacturers have integrated or invested in LiDAR?
In 2017, Detroit automaker Ford acquired autonomous driving startup Argo AI, which itself bought LiDAR startup firm Princeton Lightwave. Ford also has given financial support to LiDAR industry leader Velodyne.
In late 2017, carmaker Audi (owned by Volkswagen, the world's largest automotive company) released a partially autonomous version of its model A8 vehicle, making it the first production car to integrate a form of LiDAR.
In 2018 and 2019, respectively, Audi's Autonomous Intelligent Driving (AID) division partnered with American LiDAR vendors Luminar and Aeva for autonomous vehicles currently undergoing testing in Munich, Germany.
In December 2019, Porsche (a major shareholder of Volkswagen) announced it had invested in LiDAR vendor Aeva, with the hope of integrating its products into some Volkswagen vehicles, beginning with the ID Buzz AV minivan.
In 2018, Swedish carmaker Volvo (owned by Chinese concern Geely Holding Group) made a major investment in Luminar, which has also partnered with Japanese manufacturer Toyota on autonomous test vehicles.
Aurora, a self-driving car startup funded by online retailer Amazon, acquired LiDAR vendor Blackmore in 2019.
BMW has said it aims to launch self-driving cars in 2021 with LiDAR systems supplied by Isreal's Innoviz.
Waymo, the Alphabet subsidiary with its own LiDAR hardware, has hinted strongly that one of its goals is to ultimately sell self-driving passenger vehicles to the public. Still, so far, it has not done so. In the meantime, the Russian search engine company Yandex has its own autonomous vehicle project that is said to include an internal LiDAR team.
In the coming years, more automakers are expected to start ordering LiDAR systems in bulk as their vehicles meet and surpass the capabilities of the Society of Automotive Engineers (SAE) Autonomy Level 4+ prototype cars now being demonstrated and tested by Apple, Alphabet, Tesla, and others. Until then, lower-cost LiDAR components are expected to substantially enhance Advanced Driver-Assistance Systems (ADAS) functionality in new production cars.