At the end of March, Lotus Motors released its latest pure electric model, the Eletre, in the UK, and four lidars also equaled the Great Wall Mech Dragon, which shouted the slogan "Don't talk below 4". Two weeks later, the new electric vehicle brand Jidu Automobile unveiled the design details of the front part of its first car robot concept car, and from the information currently exposed, Jidu Automobile integrated two lidars on the cabin cover, which immediately caused a hot discussion in the car circle. I have to say that lidar in 2022 is really busy! First the number of volumes, now all start the volume layout position! Today we will have a good talk about the key sensor in the era of "new four modernizations" - lidar.
■ "LiDAR": the only way for high-level assisted driving
Before talking about the role of "lidar", we need to understand the classification of "assisted driving". According to the ministry of industry and information technology's "Automobile Driving Automation Classification" regulations, which began to be implemented nationwide on January 1, 2021, the total classification of driving automation in mainland China is currently 6 levels, and the order from low to high is 0 to 5. The various levels of standards, specifically, can be interpreted like this:
L0: Continuous lateral and longitudinal operation by the driver entirely.
L1: In most scenarios, the driver is still required to continuously operate horizontally and vertically, with functions such as adaptive cruise, traffic congestion assistance, traffic speed limit recognition, lane keeping, and lane departure warning.
L2: The driver must always be in control of driving, but the system can operate laterally and vertically in certain scenarios.
L3: The driver doesn't have to monitor the system all the time, but be ready to access and take over the drive at all times. The system prompts the driver to intervene and take over in extreme scenarios.
L4: The driver is not required in the defined scene, that is, in the defined scene, fully assisted driving can be achieved.
L5: The system can automatically respond to any scenario in the driving process of the car, without the need for a driver, which is a complete sense of assisted driving.
According to the current public perception of "assisted driving classification" in most countries in the world, L4 level and below are essentially "assisted driving", and only L5 level is called true "automatic driving". In order to complete the L3 and L4 levels, which are more "high-level assisted driving" that are free from driver interference, the choice of sensors is crucial.
At present, the mainstream sensor products used in environmental perception on the market mainly include four categories: cameras, millimeter wave radar, ultrasonic radar and lidar. Overall, the camera has poor visual effect in the case of backlight or complex light and shadow, poor recognition effect of millimeter-wave radar on static objects, limited measurement distance of ultrasonic radar and susceptibility to bad weather, so it is flawed to rely on cameras or millimeter-wave radar schemes alone to achieve intelligent driving. At this time, the emergence of lidar fills the "short board" of the above sensors.
Simplified diagram of how lidar works
In terms of working principle, lidar is very similar to "radar". Lidar takes the laser as the signal source, the pulsed laser emitted by the laser, after contact with the target, causes scattering, and part of the light wave will be reflected to the receiver of the lidar, calculated according to the laser ranging principle, the distance from the laser mine to the target point is obtained. The pulsed laser continuously scans the target at an ultra-high frequency, and it can quickly reconstruct the three-dimensional model of the target and various piece data, establish a three-dimensional point cloud map, and draw an environmental map to achieve the purpose of environmental perception.
360-degree scanning imaging of lidar
The core characteristics of lidar in assisted driving can be summarized as three-dimensional environmental perception, high resolution, and strong anti-interference ability.
Three-dimensional environmental perception: LiDAR emits a large number of laser beams to the surrounding environment in a short period of time, not only by measuring the time difference of the laser signal to determine the distance of the object, but also by scanning horizontally or to the air, and obtaining signals from different pitch angles to obtain accurate three-dimensional information of the measured object.
High resolution: The angular resolution of lidar is not less than 0.1mard, that is to say, it can distinguish two targets that are 0.3 meters apart at a distance of 3000 meters; multiple targets can be tracked at the same time, the distance resolution can reach 0.1mard, the speed resolution can reach within 10m/s, due to the high laser frequency and short wavelength, so extremely high angle, distance and speed resolution can be obtained. Such high speeds and distance resolution mean that lidar can use distance Doppler imaging technology to obtain very sharp images.
Strong anti-jamming ability: Unlike the situation that microwave millimeter-wave radar is susceptible to electromagnetic waves widely existing in nature, there are not many signal sources in nature that can interfere with lidar, so lidar has a strong ability to resist active interference and can work "all-weather".
Schematic diagram of the perceptual range of each type of sensor
In the era of intelligent cars, single-soldier combat is not the optimal solution, and multi-sensor mutual "sensing redundancy" is the future trend. With the gradual evolution of automatic driving, the amount of data in the perception layer increases exponentially, and weak perception will put forward higher requirements for the performance and computing power of the chip, increasing the difficulty of implementation. The principles and functions of different sensors are different, and it is difficult to replace each other by exerting their respective advantages in different scenarios.
The smart cars of the future can be seen as "mobile sensor platforms" that will be equipped with a large number of sensors. As assisted driving continues to advance from L2 to L3 and above, and its safety requirements are getting higher and higher, lidar plays an increasingly important role in "advanced assisted driving" with its high accuracy, long detection distance and ability to model 3D environments.
■ Of the lidars on the market, only two are mainstream
If lidar is classified according to the scanning method, there are currently three categories: "mechanical lidar", "semi-solid-state lidar" and "solid-state lidar". Among them, "mechanical lidar" is the most commonly used, "solid-state lidar" is the direction of vigorous development of the industry in the future, and "semi-solid-state lidar" is a compromise between mechanical and pure solid-state, which belongs to the main force of mass production and loading at this stage.
Mechanical-mechanical rotary lidar
Although 2021 is the first year of the outbreak of lidar, major car companies have opened an "arms race" of mass-produced sensors, but the application of lidar in the car is not new. The kind of self-driving test vehicle that we occasionally encounter on the road, the thing on the roof of the car is lidar. This is also the earliest appearance of vehicle-mounted lidar - "mechanical lidar", because the technology is relatively mature, so most driverless companies use it to do tests related to automatic driving.
How it works: The transmit and receive modules are electrically rotated 360 degrees by the motor. Multiple sets of laser wiring beams are arranged in the vertical direction, and the transmitter module emits laser lines at a certain frequency, and dynamically scans by constantly rotating the emitter head.
Baidu Apollo autonomous driving project test vehicles
Advantages: Mechanical lidar as the earliest loading product, the technology has been relatively mature, because it is controlled by the motor rotation, so it can maintain a stable speed for a long time, and the speed of each scan is linear. And because of the "standing high", mechanical lidar can reconstruct the surrounding environment with high accuracy and clarity and stability.
Disadvantages: Although the technology is mature, because of the large number of laser transceiver module wiring harnesses inside it, and the need for complex manual tuning, the manufacturing cycle is long, so the cost is not low, and the reliability is poor, resulting in low mass production. Secondly, mechanical lidar is too large and consumer acceptance is not high. Finally, its life expectancy is about 1000h ~ 3000h, and the requirements of automakers are at least 13000h, which also determines that it is difficult to go to the C-end market.
Waymo self-driving taxi overhead mechanical lidar
At present, the representative manufacturers of mechanical lidar overseas are Velodyne, Waymo, Valeo, Ouster, and domestic sagittarius Juchuang, Hesai Technology, Radium Intelligence, Beike Tianhua and so on. Velodyne's representative products include HDL-64, HDL-32, VLP-16, etc., with prices ranging from $0.4 million to $80,000. Google's self-driving car's 64-line lidar came from Velodyne, which was priced at $70,000 at the time. The high cost also determines that it is currently mainly used in the development of autonomous driving technology, such as Baidu Robotaxi, Google's driverless test fleet, and there is no application in the vehicle-grade front-loading mass production market.
Semi-solid-MEMS lidar
MeMS, the full name of Micro-Electro-Mechanical System(microelectromechanical system), is the mechanical structure of the original lidar through microelectronic technology integrated into the silicon chip. Essentially, MEMS lidar does not completely eliminate the mechanical structure, so it is a semi-solid-state lidar.
MEMS LiDAR Micro-Galvanometer Module
How it works: MEMS integrates a very compact micro-galvanometer on the silicon-based chip, and its core structure is a small cantilever beam - by controlling the tiny mirror translation and torsional reciprocating motion, the laser tube is reflected to different angles to complete the scanning, while the laser generator itself is fixed.
Sagitar Mems solid-state lidar RS-LiDAR-M1 production version
Advantages: MEMS lidar because of the removal of bulky "rotating motor" and "scanning mirror" and other mechanical motion devices, the removal of metal mechanical structure components, while equipped with millimeter-level micro-galvanometer, which greatly reduces the size of MEMS lidar, compared with the traditional optical scanning mirror, in terms of optical, mechanical performance and power consumption more prominent performance. Secondly, thanks to the reduction in the number of laser transceiver units and the possibility of price reduction of the silicon-based materials used in the overall structure of the MEMS galvanometer, the overall cost of MEMS lidar is expected to be further reduced.
Disadvantages: MEMS lidar's "micro-galvanometer" is a vibration-sensitive device, while the silicon-based MEMS cantilever beam structure is very fragile, external vibration or shock is very easy to directly cause its fracture, the on-board environment is easy to affect its service life and working stability. Secondly, the limited vibration angle of MEMS leads to a relatively small field of view (less than 120 degrees), and at the same time limited by the mirror size of the MEMS micro-galvanometer, the effective detection distance of traditional MEMS technology is only 50 meters, and the FOV angle can only reach 30 degrees, which is mostly used for close blindness or forward detection.
Signing ceremony between Sagitar Juchuang and its partners
At present, because the meMS upstream supply chain is relatively mature, such as Luminar's MEMS semi-solid-state lidar has reduced the manufacturing cost to 500-1000 US dollars, making large-scale mass production possible. In China, Sagitar Juchuang has established cooperation with 11 car companies such as GAC Eian, Weima and JiKr, and its product "RS-LiDAR-M1" has begun to be shipped in batches in December 2020, becoming the world's first vehicle-grade MEMS lidar delivered in batches. Overseas, Luminar has more than 50 industry partners around the world, including Volvo, SAIC Feifan Automobile, Xiaoma Zhixing and so on.
Semi-solid-rotor lidar
The difference between rotary mirror lidar and MEMS lidar is that the former's scanning mirror rotates around the center of the circle, while the latter vibrates up and down around a certain diameter. In contrast, rotary mirror lidar consumes less power and is less difficult to dissipate heat, so it is easier to have higher reliability.
Valeo is used in the rotorcraft lidar of the Audi A8
How it works: Unlike mems micro-galvanometers in the form of translation and twisting, the rotating mirror is a reflective mirror that constantly rotates around the center of the circle, thus achieving laser scanning. In the rotary mirror scheme, there are also two technical routes: a scanning mirror (one-dimensional rotating mirror) and a vertical and horizontal two-sided scanning mirror (two-dimensional rotating mirror). The number of one-dimensional mirror beams is the same as that of laser generators, while the two-dimensional mirrors can achieve more equivalent wiring harnesses, which has advantages in integration difficulty and cost control.
The ideal L9 will be equipped with Hesai Technology AT128 rotary mirror lidar
Advantages: The laser transmitting and receiving device of rotary mirror lidar is fixed, so even if there is a "rotating mechanism", the product volume can be made small, thereby reducing costs. And the rotating mechanism only has a mirror, the overall weight is light, the load of the motor bearing is small, the system runs more stable, the life is longer, and it is in line with the advantages of mass production of vehicle regulations.
Disadvantages: Because of the existence of mechanical forms such as "rotating mechanisms", it is inevitable that after long-term operation, the stability and accuracy of lidar will be affected. Secondly, the number of one-dimensional scan lines is small, and the scanning angle cannot reach 360 degrees.
The Valeo Scala 1 mirror lidar on board the Audi A8
From the application point of view, tier1 suppliers with vehicle-grade mass production strength include Valeo (Scala), Radium Intelligence (CH32), Innovusion (Falcon). In 2017, the Audi A8 was the world's first mass-produced L3-class autonomous passenger car, equipped with a lidar that Valeo and Ibeo jointly developed with a 4-line rotary scanner lidar. In 2020, the CH32 independently developed by Radium Intelligent was launched, becoming the second rotor laser radar in the world to obtain vehicle regulation level certification, and has been delivered to Dongfeng Yuexiang mass production front-loading models on a large scale. In 2022, NIO ET7 equipped with Innovusion Falcon lidar is available, which is a 1550nm solution, equivalent to 300 lines. In terms of price, Valeo Scala 2 is 900 euros (about 6500 yuan), which has fallen to the acceptable price range of car companies.
Semi-solid-prism lidar
DJI Incubation Technology (Livox) entered the lidar market, which is the use of prism scanning solutions, DJI uses its accumulated motor precision control technology and automated production line in the field of unmanned aerial vehicles, confident to overcome the problem of prism bearing or bushing life, but also for its lidar technology to build a moat.
Schematic diagram of the working of prism-type lidar
How it works: Prism-type lidar, also known as double-wedge-shaped prism-type lidar, consists of two wedge-shaped prisms inside, and the laser is deflected once after passing through the first wedge-shaped prism, and deflected again after passing through the second wedge-shaped prism. Controlling the relative speed of the two prisms controls the scanning pattern of the laser beam. Unlike the previously mentioned scan forms, the cumulative scan pattern of prism lidar is shaped like a chrysanthemum, rather than a row of point clouds. The advantage of this is that as long as the relative speed is properly controlled, scanning for a long time in the same location can cover almost the entire area.
DJI Livox prism lidar: Horiz (left) and Horizon (right)
Advantages: First, the design reduces the number of lines emitted and received by the laser to achieve a higher number of lines within a frame, which in turn reduces the complexity of focusing and calibration, so that productivity is greatly improved, and the cost of prism is greatly reduced compared to traditional mechanical lidar. Second, as long as the scanning time is long enough, you can get extremely accurate point cloud and environment modeling, with almost no upper limit of resolution, and can achieve nearly 100% field of view coverage.
Disadvantages: Prism-type lidar FOV is relatively small, and the scanning point in the center of the field of view is very dense, and the scanning point of the field of view edge of the radar is relatively sparse, and there will be a problem of low resolution in the short time when the radar is started. For high-speed moving cars, there is obviously no long-term scanning, but higher accuracy and longer detection distances can be achieved by increasing the laser wiring beam and power, but the mechanical structure is also relatively more complex, the volume makes the first two more difficult to control, and there is a risk of wear and tear of bearings or bushings.
The DJI Livox prism-type lidar on the Xiaopeng P5
From the perspective of vehicle regulation level application, Xiaopeng P5 is equipped with 2 DJI Livox vehicle specification grade prism lidar, and DJI Livox has also obtained the fixed point of FAW Jiefang mass production project. Dense point clouds for a single prism-style central area. With the relatively sparse point clouds on both sides, Xiaopeng P5 chose to deploy 2 lidars in front of the car, and the ultra-wide point cloud vision in the front was raised to 180 degrees, improving the ability to cope with complex road conditions such as vehicle congestion and intersection turns in the vicinity.
Solid-state-OPA lidar
In view of the characteristics of vehicle-grade equipment that needs to work continuously in continuous vibration, high and low temperature, high humidity and high salt, solid-state lidar has become a more feasible development direction. Friends who like the military should have heard of the phased array radar carried on military aircraft and warships, and the OPA optical phased array lidar uses a similar principle and moves it to the end of the car.
Schematic diagram of the working principle of OPA optical phased array
Working principle: Phased array radar emits electromagnetic waves, OPA lidar emits light, and light and electromagnetic waves also show wave characteristics, so the principle is the same. Interference phenomenon will occur between waves and waves, by controlling the current phase of each array element of the phased array radar plane array, the use of phase difference can make the wave sources in different positions interfere, so as to point in a specific direction, reciprocating control can achieve scanning effect.
Quanergy showcases the industry's first OPA-based 100-meter range solid-state lidar
Advantages: OPA lidar transmitters use pure solid-state devices, there is no need to move the mechanical structure, so the durability performance is more outstanding; although the mechanical scanning structure is omitted, it can achieve similar mechanical panoramic scanning, and at the same time can be made smaller in volume, and the cost after mass production is expected to be greatly reduced.
Disadvantages: OPA lidar on laser debugging, signal processing of the computing power requirements are very large, at the same time, it also requires that the array unit size must not be greater than half a wavelength, so each device size is only about 500nm, the requirements for materials and processes are extremely harsh, due to high technical difficulty, the upstream industry chain is not mature, resulting in OPA program in the short term is difficult to car specification level mass production, there are few Focus on the development of OPA lidar Tier1 suppliers.
Louay Eldada presents quanergy S3
At the application level, there are currently no cases of vehicle-level mass production, and the representative company of OPA solution is Quanergy. In August 2021, Quanergy completed a driving test demonstration of its OPA Guda lidar series S3 series. The test results show that the S3 series of solid-state lidar can provide more than 100,000 hours of mean time between failures (MTBF) and achieve 100 meters of detection performance in full light, with a target price of $500 after mass production.
Solid-state-FLASH lidar
Due to its simple structure, Flash flash lidar is the most mainstream technical solution for pure solid-state lidar at present. However, due to the emission of a large area of laser in a short period of time, it will be greatly affected in the detection accuracy and detection distance, mainly used for slower-speed unmanned vehicles, such as unmanned takeaway vehicles, unmanned logistics vehicles, etc., in the automatic driving solutions with lower detection distance requirements.
Schematic diagram of flash lidar working
Working principle: Flash lidar principle is similar to flash, using a camera-like mode, each pixel in the photosensitive element can record the time information of the photon flying out, running directly emitting a large piece of laser covering the detection area, and then calculating the corresponding distance information of each pixel by the high-sensitivity receiver array, so as to complete the drawing of the surrounding environment.
Schematic diagram of FLASH lidar imaging
Advantages: The biggest advantage of FLASH lidar is that it can achieve global imaging at one time to complete the detection, and the imaging speed is fast. Small size, easy to install, easy to integrate into the overall appearance of the car. The design is simple, with few components and low cost. The signal processing circuit is simple, consumes less computing resources, and the overall cost is low. The refresh rate can be as high as 3MHz, which is 100,000 times that of traditional cameras, and the real-time performance is good, so it is easy to pass the car.
Disadvantages: However, the single point area of flash laser is larger than that of scanning laser single point, so its power density is lower, which in turn affects the detection accuracy and detection distance (less than 50 meters). To improve its performance, a more powerful laser, or a more advanced laser emission array, is required to have the light emitting unit turn on and light up in a certain mode to achieve the effect of the scanner.
Receiver module for Ibeo FLASH lidar
In order to overcome the limitation of detection distance, IBM LiDAR representative manufacturers Ibeo and LedderTech began to innovate in laser transceiver modules. Ibeo, the originator of car-grade lidar, launched single-photon lidar in one step, which Ibeo calls the Focal Plane Array focal plane, which can actually be classified as Flash LiDAR. On August 27, 2019, Great Wall Motors and German lidar manufacturer Ibeo officially signed a strategic cooperation agreement on lidar technology, and the product basis of the tripartite cooperation is ibeonEXT Generic 4D Solid State LiDAR. In the long run, flash lidar chip degree is high, large-scale mass production after the probability of low cost, with the development of technology, FLASH lidar is expected to become the mainstream of the technical solution.
■ Lidar common layout
Where should lidar be installed in the vehicle? This is a topic that all car companies are currently exploring. For aesthetics and practical application needs, car companies have different considerations, from the current release of the model, the forward main lidar is mainly deployed in two areas: roof and bumper / middle net, but there are also lidar arranged on the front hood and body side.
On this issue, some time ago, the industry also carried out relevant discussions. On April 18, the details of the head of Jidu Automobile's first concept car were exposed, equipped with two lidars, and it is reported that when the lidar does not work, it can also be retracted to the inside of the hood. In this regard, Li Xiang, the founder of Ideal Auto, expressed his views on the relevant Weibo: "Put one on the roof, and put two on the cover or bumper, there is no difference in performance, and even the single performance on the top of the head will be better." However, in terms of pedestrian collisions, repair costs, and vibration control (linking the main body), the roof is optimal. The only problem is that the shape of the roof lidar will need to be adapted, because it is too much like a mobile suit - Gundam."
In this regard, Xia Yiping, CEO of Jidu Automobile, replied below the comment, "There is still a difference, 120 degrees OFV and 180 degrees of FOV are still not the same, and the solved corner case is not the same, so there is still a difference in the ability and experience of the product and even the safety."
On April 22, He Xiaopeng, CEO of Xiaopeng Motors, also released his views in the circle of friends, saying that "two lidars are definitely far better than one in terms of performance, and the difference can be seen in the future in urban NGP and robustness." On the question of where to put it, He Xiaopeng also struggled, "We first wanted to put two on the roof of the car, but no matter how designed, they are similar to the ears of rabbits, and then we gave up."
What are the different effects of the different locations, so we have made a statistics and summary of the models equipped with lidar on the market.
Roof: Maximize the benefits of a single radar
Only the layout on the roof is a popular practice in the market at present, and for models equipped with only 1 lidar, placing it on the roof can maximize the use of it. Judging from the current loading situation, the shape of the lidar installed above the roof generally has the characteristics of long and flat.
The NIO ET7 is equipped with The Innovusion Falcon lidar on the roof
The Ideal L9 is equipped with Hesai Technology AT128, the Innocent Falcon equipped with Weilai ET7 and the Luminar Iris equipped with Feifan R7 all have a horizontal FOV of 120 degrees, while the vertical FOV is 30 degrees, the ideal L9 is 25.4 degrees, and the Feifan R7 is 26 degrees. As the only lidar in the car, they can maximize their role on the roof.
The Finn R7 is equipped with Luminar Iris lidar on the roof
Advantages: It can be seen that the lidar in the table is basically based on semi-solid-state solutions, and it cannot reach the mechanical 360-degree level FOV, so "want to go to the next level", the roof layout of lidar can bring better detection effects, and it is not easy to be obscured by objects. At the same time, the layout on the roof can avoid the impact of most of the scratch collision accidents, and put the most expensive sensors in the safest place, which is undoubtedly an insurance practice.
Disadvantages: Mainly reflected in the aesthetics of the whole vehicle, not all consumers can accept this shape. Secondly, the rain and snow weather will be greatly disturbed, and the prominent part will also bring some NVH optimization problems. At the same time, although the higher the field of view, the better, the larger the blind spot of the ground line, so the roof layout needs to be fine-tuned to the lower edge of the field of view (FOV) and the inclination angle of the roof cover. Finally, summer sun exposure may cause the temperature of the element to be too high, although the lidar can work within 80 ° C, but in the long run, it is bound to accelerate the aging speed of the lidar.
Front: A favorite of traditional giants
Placing lidar on the front of the car seems to be the favorite way of traditional manufacturers, and BBA's own flagship models have placed lidar here. In the traditional factory that began to lay out automatic driving very early, lidar, like millimeter-wave radar and ultrasonic radar, is also a kind of perception radar, so it is not inappropriate to put it in the position where these old predecessors stayed, but the advantages and disadvantages of such a layout are obvious.
The new Mercedes-Benz S-Class integrates the Valeo Scala 2 into the mid-mesh
It can be seen that the two major factories of Mercedes-Benz Audi have been taken down by Valeo, and Valeo's first generation scala has given the Audi A8, which has a vertical FOV of 3.2 degrees and a horizontal FOV of 145 degrees, but the detection distance is only 80 meters. The new Mercedes-Benz S-Class is the second generation of Scala, and the upgraded product has also helped Mercedes-Benz successfully obtain the world's first conditional autonomous driving (SAE-L3 level) system international certification. Ibeo, the only supplier of flash solid-state lidar, chose to cooperate with Great Wall WEY.
Mercedes-Benz EQS also integrates valeo Scala 2 into the mid-grid
Advantages: As everyone can clearly feel, the biggest advantage of the lidar layout in the front of the car is that there is no sense of existence, and the design of the whole vehicle does not need to compromise for it. At the same time, the scheme of arranging multiple radars can make up for the shortcomings of most semi-solid-state radar FOVs.
Disadvantages: Most of the bumpers of today's models on the market are made of plastic, although it can reduce the damage to pedestrians in collisions, but it also puts lidar in a "dangerous" situation, and before the scale cost of lidar comes down, minor collisions may cause owners to pay high maintenance costs. Secondly, the lower position of the lidar arrangement is easier to be obscured by dirt, affecting the detection effect, but now some car companies have begun to equip the lidar with "cleaning devices", so this is not a fatal injury in the future.
Body side: Multi-radar blindness repair method
Compared with the first two schemes, the layout of multiple lidars is more inclined to supplement the "side view capability", in addition to the roof or front radar responsible for front view, car companies will generally arrange radar on the side of the body to form a sensing closed loop close to the full field of view. It can be seen that all of Huawei's current lidar layouts are "bumper + body side" schemes. In the case of sufficient funds, the roof layout of a forward long-range lidar, the body installation of multiple short- and medium-range angle lidar is also a more reasonable and safe solution.
The WM M7 is equipped with three lidars
The WM M7 is equipped with three M1 lidars from Sagitar Juchuang, which are located on both sides of the roof and fenders, and can achieve a horizontal detection range of up to 330 degrees. The salon car, which high-profile declares "don't talk below 4", is equipped with 4 lidars for its mech dragon. Also using the four-element lidar scheme is the Lotus ELETRE released in March, which has two "telescopic" 128-line liDAR.
Lidar located above the front arch of the Avita 11
Advantages: From the coverage point of view, in the roof / front of the installation of a forward lidar program, in the four corners of the vehicle to supplement 2-4 short-range lidar is the most comprehensive approach, basically can do 360 degrees of horizontal FOV, while in the vertical FOV is also greatly improved, in the L3 level of driving assistance to L4 or higher level of automatic driving in the process of moving forward, multi-lidar program will gradually become the mainstream.
Disadvantages: high yield is naturally accompanied by high risk, the shortcomings of the above layout will almost always be involved in the multi-lidar solution, whether it is the thermal management of the roof radar, or the easy collision of the front radar, the biggest disadvantage of the multi-radar can be said to be the cost, whether it is pre-sales or after-sales, the multi-lidar solution means that more money needs to be paid.
Hatch: A different approach to set size
In addition to the above three mainstream layout methods, Jidu has created a precedent for the layout of the cabin canopy, which is a very novel layout method and will inevitably cause many views in the industry.
Hesai Technology AT128 lidar on jidu automotive robot
In the previous communication meeting, we confirmed that Jidu will use Hesai Technology's AT128 lidar, which is the same parameter as the one on the roof of the ideal L9. From the official picture, we can see that the two lidars are located at the left and right ends above the hood, similar to the retro "jump light" design.
What the lidar looks like after it is lowered
Advantages: The advantages of the set degree dual lidar scheme are mainly reflected in the horizontal FOV, which is 120 degrees of AT128 lidar level FOV of Hesai Technology, and the two can achieve 180 degrees of coverage. In the overlapping area of 60 degrees directly in front, due to the high density of point clouds, its recognition ability is also stronger and the accuracy is higher. Dual lidars are mutually securely redundant and more reliable than single lidar solutions. In driving scenes such as "ghost probes" and left and right obstructions, the ability to recognize pedestrians or obstacles crossing the left and right is stronger. At the same time, the "lifting" lidar can be retracted when parked, avoiding long-term parking wind and sun and prolonging the service life.
Disadvantages: As Li xiang said, the layout mode of the cabin cover is not very friendly for the safety of pedestrian collisions, and the position of the edge of the lidar shell, the hard protrusions during the collision are likely to cause more damage to pedestrians. There is now a "pedestrian protection leg type test", and the main standard of this test ("GB24550-XXXX Car Collision Protection Standard for Pedestrians") will become a mandatory national standard on July 1, 2024. However, this point is expected to be optimized through OTA in the later stage, and the lidar will be lowered at the moment of collision, protecting both pedestrians and lidar.
■ Bang reviews
Although lidar is transitioning from semi-solid to solid and has not yet been applied on a large scale, car companies have begun to focus on the number of "inner volumes". We need to make it clear that lidar is only one part of the assisted driving perception hardware, and the realization of "high-level assisted driving" needs to be combined with other sensors and algorithm platforms, rather than blindly pursuing quantity. The layout of lidar depends on how to balance perception, aesthetics, cost, and safety.
Looking back at history, Tesla has gone from Mobileye to self-research, the technology iteration speed is fast, and its functions are becoming more and more practical and easy to use. Therefore, we are more willing to look at it with the first principle of this "inner volume" of lidar that has been set off, and the essence of this "inner volume" is the controversy of a hundred schools of thought in the emerging period of technology, which will eventually promote the historical process of "high-level assisted driving". The mass production of lidar is a long road, and the "inner volume" of quantity and layout is only temporary, and we still have a long way to go before it really enters the lives of the people. As the saying goes in The Legend of the Galactic Heroes:
Our journey is a sea of stars.
![Lexmark A4 format medium-sized workgroup laser printing series new products are available[fig]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)