Shenzhen-Bengaluru, October 09’2020 - The autonomous driving industry has always seen some kind of narrative around LiDAR sensors both from the positive and negative perspective. However, there is no doubt that the technology is flourishing and been used by majority of OEMs and Tier 1s as a key 3D perception and sensing for not just autonomous vehicles but also in the mobility infrastructure.
Eva Sharma, Consultant, ACES at M14 Intelligence exchanged dialogue with Dr. LeiLei Shinohara, the Vice President and Co-partner at RoboSense on some of the most crucial questions that exists in the LiDAR’s mass production for autonomous driving application. We appreciate the honest views of Dr. LeiLei on each topic and congratulate him for the success he has achieved over the years. Below is a snapshot of the exhaustive discussion.
What hurdles the Covid-19 breakdown has brought in the way of RoboSense? What challenges did you face in achieving your 2020 milestones?
All types of businesses in all major countries have been impacted to a certain extent during the pandemic breakdown. For RoboSense, there are majorly two aspects that received an impact from the COVID outbreak.
The first and the most important aspect is the impact on the global customer base especially from the European region and some of the American countries. Quite a lot of our customers have stopped their operations and autonomous vehicle testing. Though the United States has recently resumed the AV testing, the overall number of the testing fleet has been reduced. Except for Asia, all other regions have slowed down their autonomous driving and RoboTaxi activities. Also, most of the countries have witnessed a significant reduction in vehicle production and sales, which had directly impacted our market as well. However, China and most of the other Asian countries have been less impacted or we can say showed early recovery. Particularly in China, the months of February, March, and April were highly crucial as the market was under lockdown; however, it showed amazingly fast recovery immediately after April.
During this COVID, the Chinese government took quite a lot of efforts to avoid extreme economic losses by deploying unmanned vehicles to deliver food, essential items, and packages. Owing to this, the country has witnessed a speedy development in the autonomous commercial vehicle industry.
The second aspect is the decline of the overall automotive industry. It is not that just this year but from the last year itself, the automotive industry has gone down and seen a continuous fall in sales. COVID has further aggravated the situation. In China, the 3rd quarter of 2020 has witnessed a speedy recovery in the automotive industry; although, the auto sales had fallen significantly in Q1 and Q2.
The OEMs working on autonomous driving technology have also paced up their research and development operations to ensure high-level safety. The recent accident of a Tesla Model X in California which was on Autopilot mode has also made the industry realized the importance of LiDAR sensing for autonomous vehicles. The recent premiere of the new Mercedes Benz S Class revealed the use of LiDAR in the vehicle, so is Volvo, Toyota Lexus, and Lucid’s new electric vehicles are going to use LiDAR in their sensor suite. We have received interests of lots of OEMs and other customers for our LiDAR products to incorporate in their upcoming autonomous cars. Therefore, I could say that COVID has in fact speed up our LiDAR development and created good opportunities for RoboSense.
Considering the geo-political situation like the US-China trade war, how do you see the effects on demand for RoboSense LiDARs from overseas?
Currently, the autonomous driving market is at a very nascent stage and we (RoboSense) provide LiDAR samples for testing AVs which are not in big volumes. However, in the later stage, there might be a certain level of impact on our market due to these geo-political conditions.
We do have a customer base in North America who are still procuring our LiDAR samples for testing. However, the volume is quite less which is also because of the COVID -19. The entire North American market has shrunken quite a lot. This is not only for RoboSense, but even the other suppliers from European and American countries are following similar trend.
Talking about the US-China trade war, right now for the sample phase, we do not see any change in the demand. I strongly believe that even later in the mass production program, RoboSense would receive an incredibly good market demand as we are confident about our price and quality of the product.
Do you think the autonomous driving solution will be successful without LiDAR? Will LiDARs find unique position in the autonomous driving sensor suite?
(Background - If LiDAR sensors both short range and long range are included in the sensor suit of autonomous vehicles, the overall autonomy package cost increases. This is the major reason behind the companies like Tesla and Xpeng to avoid the use of LiDAR sensor and focus on cameras and radars)
Currently, as it is just a beginning of autonomous driving technology, the automotive leaders are slowly getting towards the launch of so-called autonomous vehicles. Level 2 functionality is already existing in few of the car models and OEMs are promoting the advanced generation of level 2 features as level 2.5 and not level 3, except for Tesla and Audi. The major reason behind this is that the OEMs are defining their own autonomy features with certain conditions applied and do not want to name it as level 3. Level 3 according to SAE standards is conditional automation with eye-off the road, which I believe needs quite strong perception models and sensor redundancy.
We already have an example of recent Tesla accident that claimed to be level 3 equipped automation and even Traffic Jam pilot in Audi A8 is not being considered as Level 3 automation in the U.S. and company is no longer adding this feature in the current generation cars. Whichever OEM is launching the Level 3 vehicles without LiDAR is putting lots of limitations on the system. For example, the limitations could be such that the vehicle would not support automation during night, the driver must take control on sharp turns, and so on.
Couple of our customers are working on level 3 automation; however, there are lots of customers who are planning to launch the fully autonomous driving capability vehicles with automatic lane changing feature. Such customers have realized the importance of LiDAR and have considered it as one of the inevitable sensors for vehicle autonomy. As I mentioned in the above point that OEMs such as Audi, BMW, Volvo, and Toyota have positioned LiDAR in their autonomy platform.
Also, to have the ability of level 3 autonomy and make the vehicle features better, the most important thing is to improve safety on detection and safety on perception. For example, the passenger/driver losses trust on the autonomous system and gets frustrated if the system is signalling/alerting him/her every minute to take control of the car, which is not expected from a system that is called as autonomous.
This interruption gives a bad user experience and the driver uses its energy focusing on the road to make sure if the car is heading safely without any strange movements. The user will no longer trust such systems and will prefer the manned driving over the unmanned. However, if LiDAR is used as a part of such systems, it will certainly improve the system perception, performance, and the user experience.
There are many LiDAR technologies in the market like flash, OPA, mechanical scanning, and MEMS solid-state. On which technology RoboSense is focusing and which technology according to you is preferred by OEMs?
Yes, there are different LIDAR technologies as you mentioned. As far as RoboSense is concerned, in automotive-grade, we are focusing on MEMS-based Solid-State LiDAR technology using 905nm laser diode because it is the most manufacturable technology and can achieve mass production.
There are some companies which are using 1550nm wavelength laser, but 905nm is most widely used in MEMS and is a stable LiDAR technology. Although the 905nm cannot match the improved sensor performance of 1550nm, it will take quite a couple of years or more to get 1550nm at an affordable price with a reduced power consumption rate. Hence, from my perspective, in the current market, the MEMS-based 905nm technology is the best fit for long-range detection.
Also, mechanical LiDARs are being used for multiple years already and are the most stable technology. At present, RoboSense has a full range of mechanical products such as 128 lasers, 80 lasers, 32 lasers, 16 lasers. With extremely high performance and stable reliability, we have won a large number of customers in RoboTaxi, RoboTruck, robots, ShuttleBus, and other fields. During the COVID-19 outbreak in China, more than 70% of the robots were equipped with RoboSense LiDAR; however, in terms of passenger cars, it has certain issues especially for automotive grade like low shock resistance and large size and high price.
As you mentioned about the flash LiDAR, they are available from Continental, TI, and other companies who have released these LiDARs for automotive grade. RoboSense is also working on flash LiDAR; however, right now we are focusing to use the flash technology for short-range detection. For example, to detect the vehicles for automated parking and low-speed functionalities.
We are working aggressively on LiDAR technology and we believe that in the short-to-mid-term i.e. 3 to 5 years, for long-range detection MEMEs solid-state LiDARs and for short range detection flash LiDARs have good market potential. Also, the mechanical LiDARs still have a market for Robotaxi and Robotic application.
What is the timeline for RoboSense to start mass production? What is your target pricing for higher volumes of LiDARs?
As LiDAR technology is in the budding stage of the market, most of the customers do not have the requirements for big volumes; however, we have customers who are ready to start their SOP and mass production from next year. Our mass production could predominantly start for customers in next year. Also, for the new energy companies and automotive leaders, it is the first generation of their autonomous systems to be launched in the market. These first-generation systems are majorly focused on collecting enough data and feedback from the users about the system and experiences. We expect that the release of second-generation systems i.e. from 2023 or 2024 could bring a heavy volume demand for our LiDARs.
Regarding the pricing part, at least in the near future like 2022 or 2023, the acceptable pricing for the OEM customers could be much less than 1000 USD. However, we have also received requests from most of our customers for LiDARs in the price range of 500 USD or even lower. We have also seen that the customer losses interest if the price of LiDARs is more than 1000 USD. Hence, we expect that for bigger volume deals, the price could reach somewhere below 500 USD.
About RoboSense RoboSense since 2014 is a leading provider of Smart LiDAR Sensor System solution autonomous driving passenger cars, RoboTaxi, RoboTruck, automated logistics vehicles, autonomous buses, and intelligent roads. The company’s automotive-grade solution is based on MEMS-based Solid-State LiDARs with AI-based fusion systems. Mechanical, Flash, and fusion hardware are some of the other technologies that the company has in its product portfolio. RoboSense has a created a strong hold in the industry with more than 500 patents and consistent winner of the CES Innovation Award for the past 2 years. It has a large customer base in the automotive sector and has expanded its reach by recently partnering with Alibaba’s Cainiao Network, SAIC, BAIC, and FAW are some of the company’s strategic partners and investors. The company expects robust demand for its LiDARs from all over the globe due to its low cost to performance ratio LiDAR products.
LiDAR - Market Growth, Potential, and Supplier Mapping
Read in-depth analysis on LiDAR market for autonomous driving application and ADAS functions