Autonomous driving is difficult to get on the road: it turns out that LiDAR is holding it back

　　In the past two years, the concept of autonomous driving has swept the automotive industry, software technology has advanced by leaps and bounds, and mileage has repeatedly set new records. However, the industry's rapid progress is encountering hardware bottlenecks - LiDAR , one of the core hardware of the autonomous driving system, is slowing down the wheels of autonomous driving due to its high cost and limited production capacity . Let's follow the automotive electronics editor to learn more about the relevant content.

　　In the autonomous driving industry, despite controversy, the mainstream still believes that LiDAR is the most important component for achieving autonomous driving. This device actively emits laser beams, receives and calculates reflected laser signals, and draws 3D images of the surrounding environment in the form of "point clouds". Many autonomous driving test vehicles, including those from Google, Uber, and Toyota, rely heavily on LiDAR to achieve positioning on high-precision maps and distinguish between pedestrians and other vehicles. The best-performing LiDAR can achieve centimeter-level accuracy at distances of more than 100 meters.

 Autonomous driving is difficult to get on the road: it turns out that LiDAR is holding it back

　　Point cloud data generated by LiDAR

　　Most companies moving toward commercialization of super-autonomous driving believe that lidar is essential. (Except for the radical Tesla, they only use cameras and radar.) Because radar has poor detection capabilities for details, and cameras perform poorly in low light or strong light conditions. Tesla's fatal case last year was caused by Autopilot's inability to distinguish between a reflective trailer and a bright sky, causing the Model S to rear-end the trailer. Toyota Vice President Ryan Eustice said that whether lidar should be used in a more conservative and safer autonomous driving system is still "a question."

　　However, the development of autonomous driving is very fast, and this emerging industry is facing a problem: LiDAR is lagging behind. LiDAR used to be a small business, and the technology was not mature enough to be standard on millions of cars.

　　If we look at the current prototypes of self-driving cars, we can easily find the problem: LiDAR is too bulky. Google, Uber, and Toyota's test cars all have this huge, constantly rotating thing on their heads.

　　In addition to the size problem, LiDAR is also very expensive, with a single unit costing thousands or even tens of thousands of dollars. Most test vehicles are equipped with multiple LiDARs, and although the number of test vehicles on the road is small, meeting demand has become a problem. The Information reported last week that LiDAR manufacturers' production capacity issues have forced downstream companies to wait until the first half of the year to get the product.

　　That may partly explain the lawsuit Waymo filed against Uber last month, in which it said it had evidence that Anthony Levandowski, its former top engineer, stole Waymo's lidar designs before leaving to start self-driving truck company Otto.

　　When I visited Otto last year, Otto co-founder Lior Ron told me that Otto made its own sensors because there were no products on the market that could meet the needs of 18-wheel trucks. Now, Waymo says that Otto's technology is actually developed by itself, and the Waymo team has invested tens of millions of dollars in this technology, achieving a 90% reduction in lidar costs while improving performance.

　　Better lidar is a core part of Waymo's autonomous driving commercialization plan. The company has developed three lidars with different detection ranges. Waymo also said that it will license a series of important technologies including lidar to automakers in the future.

　　Waymo is not the only company spending big to solve the lidar problem. Last year, Ford and Baidu jointly invested $150 million in Velodyne, a leading lidar manufacturer. Velodyne is building a large factory in San Jose, Silicon Valley, to expand production capacity and plans to start producing lidars next year.

　　However, many in the autonomous driving industry believe that lidar needs to be redesigned to be practical. Velodyne is currently working on solid-state lidar, which allows the lidar to cover the road with laser beams without rotating. (Translator's note: Velodyne has not yet launched a fully solid-state lidar. Its product that is closer to this form is semi-solid-state lidar PUCK). Lidar working in solid-state form will be much cheaper, smaller and more durable because it has no mechanical parts that need to drive the radar to rotate.

　　Velodyne announced last year that their project had made a breakthrough that would allow the price of LiDAR to reach $50, but did not disclose when the solid-state LiDAR would be mass-produced. Another startup that focuses on solid-state LiDAR, Quanergy, which received $90 million in funding last year, announced that it will start production of solid-state LiDAR at its Massachusetts factory this year. The product will be priced at $250, but its specific performance is still unclear. Automotive parts suppliers Continental and Valeo are also developing similar technologies, but they say they will not be available on the market until two or three years later.

　　A group of automakers, including Ford and BMW, have said they hope to have self-driving cars on the road in 2021. These lidars under development will greatly affect the performance, cost, and appearance of future self-driving cars.

    The above is an introduction to the automotive electronics industry: Autonomous driving is difficult to get on the road: it turns out that lidar is holding it back. If you want to know more related information, please pay more attention to eeworld. eeworld Electronic Engineering will provide you with more complete, detailed and updated information.