From the $75,000 question to becoming ubiquitous on Chinese smart EVs, for show or for dough?
SHANGHAI — Chinese smart EV startup WM Motor announced the global premiere of the M7, billed as the brand’s first smart, all-scenario, 100% electric sedan, at an event here on October 22.
The most striking feature of the M7, the fourth model from the Shanghai-based startup based on the MAVEN Concept that was unveiled back in May 2020, is the bump atop the front windshield as well as one on each side of the car just above the wheel arch and below the A pillar.
Yep. You guessed it. Those are LiDARs.
But they are not just any LiDARs.
WM Motor claims that the M7 is the world’s first smart EV to incorporate three self-zooming, high precision, beyond-visual-range (BVR) quasi-solid-state automotive grade LiDARs, and also the first to achieve a combined horizontal detection range of 330° from those LiDARs. They use the so-called MEMS (Micro-Electro-Mechanical Systems) technology which is safer, more reliable and substantially more energy-efficient than comparable counterparts since moving parts steer only the laser beam in free space without moving any optical components. This further allows for autonomous adjustment of the LiDARs’ refresh frame rates and resolutions to conform to the various perceptive requirements needed for different scenarios such as urban areas and highways.
The three LiDARs are part of a total of 32 perception sensors including HD cameras, ultrasonic radars and millimeter wave radars fitted around the M7 dedicated to smart assisted driving.
Behind the LiDARs is Shenzhen-based RoboSense, a company that was founded only 7 years ago but is already supplying LiDARs — including the second-generation RS-LiDAR-M1 for the M7 — for a variety of applications including ADAS for private passenger vehicles, unmanned logistics vehicles, robots, RoboTaxis, RoboTrucks, RoboBuses as well as new infrastructure for smart transport. In addition to the WM Motor M7, it has officially reached a deal to supply LiDARs to the GAC AION LX, and is rumored to be the LiDAR supplier for SAIC Motor’s upcoming IM L7 as well as future variants of the recently launched Lucid Air. It also supplies to Chinese autonomous trucking startup Inceptio which provides solutions for Chinese truck makers Dongfeng and CNHTC. More production smart EVs pre-installed with RoboSense LiDARs, most of them from Chinese brands, will be announced in the coming weeks and months, according to a PR manager with the company.
RoboSense is just one of several Chinese LiDAR suppliers that are supplying automotive grade LiDARs to production smart EVs that have already launched or are being launched over the next several months: the Xpeng P5 that was officially launched in September and started deliveries in October is equipped with three LiDARs from Livox, a spinoff from leading Chinese drone maker DJI; the NIO ET7 due to start deliveries in Q1 2022 is equipped with a LiDAR from Innovusion, which was originally founded in Silicon Valley but now calls Suzhou its home; the ARCFOX Alpha S HI edition that will soon start deliveries this month is powered by three LiDARs from Huawei, so will the upcoming E11 from Chang’an’s high-end AVATR brand; and just last week, the HiPhi Z unveiled by Human Horizons due to start deliveries in the second half of 2022 is equipped with a LiDAR from Hesai Technology.
The only other prominent LiDAR supplier that has made inroads into the Chinese smart EV space is U.S.-based Luminar, which is supplying LiDARs for the ES33 electric SUV coming in 2022 from SAIC Motor’s R Auto (recently rebranded to Feifan Auto) brand. Luminar also supplies the next generation XC90 EV due out next year from Volvo Cars, which is owned by Chinese parent company Geely.
RoboSense, Livox, Innovusion, Huawei and Hesai, in no particular order, are what I call the Chinese LiDAR supplier “Big 5.” All of them are already supplying to at least one if not multiple OEM customers (not exclusive to Chinese OEMs) and production smart EV models, not to mention many of the RoboTaxi, AV, autonomous trucking and logistics companies including Baidu, DiDi, WeRide, AutoX, Pony.ai, Momenta, JD, Neolix, Zoox, Nuro, Waymo, Cruise, Kodiak, DeepRoute, Qcraft, etc. All of them are young, having been established only within the last seven years (Huawei was founded in 1987 but only started development of LiDARs in 2016 and its automotive grade LiDAR was officially unveiled just a year ago). All of them have the speed, agility and ingenuity at developing LiDARs and making them light, compact, easy to integrate and most importantly, automotive grade, production ready and cost effective. All of them supply LiDARs with some unique selling point, and all of them have one thing in common: they are all headquartered in cities with names that start with the letter “s” (see Quick Glance at Chinese LiDAR Supplier “Big 5”).
And, get this: they are just five of reportedly close to 50 Chinese companies big and small engaged in the development of LiDARs.
*Huawei officially announced its entry into the auto industry as a value added supplier in April 2019 at the Shanghai Auto Show, LiDAR development began in 2016
The LiDAR, first patented by Velodyne founder David Hall back in 2006 as part of the DARPA Grand Challenge that year, used to be a $75,000 question: the technology is nice and critical to the application of AVs but it was super expensive because it was difficult to scale due to the complexity of mechanical parts and electronics. Now, thanks in part to some reverse engineering and rapid iteration capabilities of the “Big 5,” it’s a sub-$1,000 question as it becomes a ubiquitous feature on Chinese smart EVs. And unlike the bulky, KFC bucket-style LiDARs of the early days, they have become much more compact: take the RoboSense RS-LiDAR-M1 or the Hesai AT128 for example, they fit into the palm of your hand and are only as thick as a book.
They also have some out of the world specs. The Innovusion Falcon for the NIO ET7, for example, can detect objects as far as 500 m away and 250 m detection for 10% reflective objects with super high resolution of 0.06x0.06 degree. Even pedestrians and small road debris can be detected over 200 m away. The Livox HAP on the Xpeng P5 can detect objects with 10% reflectivity 150 m away, features a horizontal FOV of 120 degrees, an angular resolution of up to 0.16x0.2 degree, a point cloud density equivalent to 144 lines and 450,000 data points generated per second. The Hesai AT128 auto-grade long-range hybrid solid-state LiDAR for the HiPhi Z claims to be the only LiDAR on the market that is capable of detecting objects with 10% reflectivity from 200 m away and generating 1.53 million data points per second under single return mode. It integrates 128 channels onto a single circuit board and features solid-state electronic scanning in the vertical direction. It also has an ultra-wide FOV of 120 degrees, resolution of up to 0.1x0.2 degree and dimension of just 137x112x47 mm (LxWxH).
So why are we seeing these LiDARs all of a sudden “pop up” from production smart EVs this year, especially those from the startups?
It’s both for show and for dough.
For starters, the battle of the smart EVs in the future will be won on “smart” rather than on “EVs,” what Xpeng refers to as the “second half of the smart EV competition.” Companies like Xpeng have put autonomous driving and full-stack in-house AV development capability at the center of their strategies. While Level 2 driver assistance features are becoming commonplace, they are still limited to certain driving scenarios due to the limitation of existing sensor setup lacking LiDAR. Now that the LiDAR is becoming cheaper, more compact and easier to integrate, companies like Xpeng and NIO are adding them onto their vehicles to add safety redundancy and an extra “pair of eyes,” so to say, so that Level 2 driver assistance can be utilized in more scenarios to eventually all scenarios. This is exactly what Xpeng is trying to do with its XPILOT software and NGP (Navigation Guided Pilot): expanding XPILOT 3.0 highway NGP (currently available) to XPILOT 3.5 city NGP (available first half of 2022) with LiDAR help, and to XPILOT 4.0 door-to-door NGP (first half of 2023).
LiDARs are also adding an extra revenue stream for the smart EV startups: many of them are offering model trims that are hardware ready, but users must pay an additional monthly or annual fee for the software upgrade (as is the case for Xpeng’s XPILOT and NIO’s NAD). And frankly, LiDAR, because of its ability to see better and detect moving objects far away better, has become a “cool” feature and a hot selling point in terms of marketing jargon: many of the aforementioned models with LiDAR claim to be the “world’s first production smart EV with LiDARs.”
For the longer-term, many of the smart EV startups have set sights on L3/L4/L5 levels of automation, and the industry consensus (except Tesla) is that LiDAR must come into play in these levels of automation. Volkswagen is already working on its Trinity project that will see a L4-capable vehicle for private purchase by mid-decade, so is Xpeng and Geely. Xpeng, in particular, is even considering launching its own RoboTaxi service using its own fleet of smart EVs with LiDARs.
From now until then, however, the mass adoption and commercialization of smart EVs with LiDARs still needs to clear regulatory as well as quality hurdles. Xpeng, for example, plans to make XPILOT 3.5 city NGP available in limited cities at first and then gradually expand nationally, a process that will take about a year. How these LiDARs perform under real word everyday driving conditions and whether their quality will hold up remain questionable.
Companies like autonomous trucking startup TuSimple is taking a slightly different approach: unlike competitors who use LiDAR as their primary sensor, TuSimple has taken a unique path in using high-definition cameras as its primary sensor technology. And by integrating LiDAR and RADAR as complementary sensors, TuSimple’s L4 autonomous driving system is capable of identifying objects up to 1,000 m or 30 s ahead while operating at highway speed and operating safely and reliably in the dark and in various weather conditions.
In a sense, LiDAR isn’t everything, but there is no denying that it is becoming a key differentiator and enabler of higher levels of automated driving covering a wider array of driving scenarios, and that 2021 is the year of the (Chinese) LiDAR as it is finding its way onto smart EVs for private purchase.
