Yole Group, including Yole SystemPlus and Yole Intelligence, are recognised for their expertise in the analysis of markets, technology developments, and supply chains, as well as the strategies of key participants in the semiconductor; photonics, and electronics sectors. In the automotive field, particular areas of interest are lighting and lidar applications.
Today we interview Pierrick Boulay, a lighting and display senior technology and market analyst in Yole Intelligence’s Photonics and Sensing division. Boulay holds a master’s degree in Electronics from ESEO at Angers, France. He carries out out technical; economic, and marketing analyses of lighting systems in general, with special emphasis on solid-state lighting. In addition, he leads the automotive activities within the company. He has authored several reports and custom analyses on topics such as vehicle lighting; lidar; sensing for ADAS vehicles, and VCSELs. Prior to Yole Intelligence, he mostly worked in R&D departments on LED lighting applications. Boulay is a highly respected speaker at DVN events.
Yole Intelligence and Yole SystemPlus offer a vast collection of reports focused on lidar and radar throughout the year. In this context, we were interested to learn more about Yole Group’s approach and views on the future of the lidar market and technology. Mr. Boulay graciously granted us this interview:
DVN-Lidar: Lidar sensors are still expensive due to the early stage of the learning curves. From your point of view, which elements can generate significant opportunities for cost reduction in the next five years?
Pierrick Boulay: Today, the volume of lidar implemented is very small compared to other sensors such as radar or cameras, which benefit from the volume effect to lower their cost. Once lidar shipments are high enough, the volume effect will have a substantial impact on lidar’s cost. If we look at the cost of some components themselves, those that are essential are still very expensive, such as the laser source; the photodetector, and the embedded FPGA. There are only a limited number of suppliers for these components. With more competition, the cost of these components should decrease as well.
DVN-L: From a cost perspective, will there be a tipping point after which automakers will start implementing lidar sensors in bigger quantities to improve their ADAS and AV functionalities?
P.B.: The cost of a radar or camera is between 0.1× and 0.2× that of a lidar. So lidar can only be implemented today in high-end cars, the E segment. It is also linked to the willingness of automakers to achieve L3 automation. Adding a lidar is not as simple as adding another sensor to a car with its ECU; it is part of the significant change in the E/E architecture in cars evolving from a distributed architecture to a more centralised one with the end goal of data fusion. According to recent feedback, a cost between $300 and $500 could be an enabler for the implementation of lidar in C and D car segments.
DVN-L: Another interesting thought is the integration of lidar sensors with front and rear lamps because requirements for lidar and lighting show similarities in several areas. What are the pros and cons of such an approach in your view?
P.B.: There are multiple locations where lidar could be integrated. lidars were initially integrated into the front grille, but recently other positions—such as the roof; the windshield; fenders; bumpers, or headlamps have become possible. The main advantage of lidar integrated into the lamps is the location on the corners of the car, enabling a very large field of view. The protection offered by the lamp cover is a clear advantage, and front cleaning systems can be re-used on headlamps. Another advantage is linked to the wiring, as the wiring of the head or rear lamps can be used for the lidar. As wiring is one of the heaviest components in cars, this is something to consider. In terms of drawbacks, however, implementing lidar in headlamps would mean doubling the number of lidars necessary for L3, as only one lidar may be enough if placed in a central position (roof, windshield, grille). Heat issues will also have to be faced as LED modules also produce heat, and active cooling is already necessary in some cases. We also see a trend to a slim design of head or rear lamps. The integration of lidar would require the volume of the sensor to be reduced drastically, which has so far not been possible. Implementing lidar in headlamps could be possible, but only for higher levels of automation when more lidar per car would be necessary.
DVN-L: The lidar market is crowded with system and component suppliers as well as software/AI companies. Yole have mapped this ecosystem in different reports. Will this crowded field help to speed up market development of lidar sensors, do you think? Or might it have the opposite effect and be an obstacle?
P.B.: The lidar market is crowded with system suppliers, and we have counted around 60 companies targeting the automotive market. This high number of companies brings diversity in technology, whether imaging (mechanical, MEMS, flash) or ranging (ToF, FMCW, or phase shift). Everyone wants to enter the automotive market as quickly as possible to generate revenue. It is a clear technology push, and it could help with the faster market development of lidar.
DVN-L: In connection with the previous question, will industry standardisation or even government regulation be an enabler for a market breakthrough of lidar systems, or would such initiatives hinder market development?
P.B.: In our understanding, regulation cannot impose a particular technology. The ALKS regulation does not mention the type of sensor, or location, or technology, and only use cases are mentioned. Organisations like NCAP or NHTSA could reinforce the performance of specific tests in low-light conditions. But in this case lidar is not the only sensor that could be used; thermal cameras could also be used.
DVN-L: Do you expect a consolidation of suppliers in the lidar ecosystem? If consolidation happens, will technological aspects like scanning principles, range performance, and speed detection play a key role, or will there be more focus on business aspects, like coöperations and alliances or finance-like funding?
P.B.: Yes, consolidation will happen. If you make the comparison with the camera and radar markets that are quite mature, you clearly see that four or five suppliers control 75 per cent of the market. We expect a similar landscape when lidar matures. We should see more mergers between lidar suppliers, and tier-1s acquiring lidar makers. It is too soon to tell if one technology will prevail. There are still some improvements to be made at the emitter or receiver levels, and ToF lidar at 905nm is still improving. To reach long range, we have seen several lidar manufacturers using a 1,550nm architecture. More recently, we have seen interesting progress with 905nm-based lidar using a new generation of components such as VCSELs and SiPM to replace the traditional EEL (Edge Emitting Laser) and APD (Avalanche Photodiode).
DVN-L: Lidar sensors are integrated into vehicles, not as stand-alone systems, but to enable advanced ADAS and AV functionality in combination with cameras and radars—sensor fusion. Suppose this fusion progresses successfully and regulatory barriers fall. What is your expectation for L3-capable cars on the road in Europe, North America, China, and Japan in significant quantities? Do you foresee regional differences?
P.B.: European car makers were the first to implement lidar with Audi in 2018 and now Mercedes and BMW. Then others followed—Honda in Japan; GM in the US— and now many new EV automakers targeting high-end cars are implementing lidar for L3. Today there is a clear move from Chinese new EV automakers towards L3. What is interesting with these Chinese automakers is that they are implementing multiple lidars per car together with the necessary hardware to reach L3, but there is currently no regulation to enable this automation level. So, lidar is not activated yet and is only used to collect data for the automakers. As soon as the regulation is in place, automakers will be able to activate new automated driving features by OTA (over the air) updates. Regarding historical automakers, recent moves from Stellantis and Nissan have been observed to implement L3 functionalities and therefore integrate lidar. Other automakers could be in a wait-and-see mode or directly leapfrog L3 to build real autonomous cars.