In what direction will the market evolution of lidar develop?
In the past 20 years, millimeter wave radar has experienced from the single function of ACC and AEB to now integrate with the camera to become one of the main sensing components of ADAS, and the number of carriers has increased from 1 to 3 or 5, and the industry is expected to experience a similar development curve.
, and is responsible for product development for the mass production market, as well as vehicle-grade manufacturing, validation and testing.
On this basis, Continental will launch a mass-production system for autonomous driving for highway scenarios around 2024 based on the multi-sensing fusion capabilities of short/long-range lidar, cameras, and millimeter-wave radars, including 4D imaging radar.
According to the plan given by Continental, the primary target is the high-end passenger car market and the commercial vehicle autonomous driving market. Among them, in the field of commercial vehicles, AEye and Continental have reached a technical cooperation agreement with Tucson Future.
Obviously, different technical routes, different application scenarios, and different perception combinations will have a direct impact on the demand stratification of the subsequent market of lidar.
<h1 class="pgc-h-arrow-right" data-track="15" ></h1>
Public information shows that Continental has named its first long-range lidar, the HRL 131, a MEMS long-range lidar. The short-range lidar that has been mass-produced is called HFL110, which is a FLASH solid-state high-resolution lidar suitable for the L2+ and L3 markets.
The HRL 131 is a 1550nm MEMS lidar jointly developed by Continental and AEye.
As we all know, the size of the MEMS mirror largely determines the reliability of lidar. Larger mirrors also have greater inertia, producing 10x to 600x the shock and vibration torque. In addition, larger mirrors do not allow for fast and agile scanning.
AEye's unique meMS patented system design allows for a mirror smaller than 1 mm in size (extreme shock and vibration resistance), while other similar lidar schemes typically use mirrors from 3mm to 25mm, adding complexity and cost to some extent. In addition, with 1550nm lasers and precision receiver modules, a breakthrough in detection distance is achieved.
Previously, for the MEMS solution, another lidar company Luminar said in the prospectus that the range/resolution is easily limited by high noise, and the fragility of the MEMS galvanometer requires high-specification special production processes and quality management to solve, and most enterprises are difficult to completely solve in a short period of time.
But in AEye's view, MEMS itself is made of monocrystalline silicon material, which is strong and durable, resistant to material fatigue, and has high temperature and impact resistance. At the same time, there is a reflective coating in the silicon that can enhance the reflection of light. At present, the industry is seeking a breakthrough in the design innovation of MEMS.
And once you break through, you have the opportunity to open up new spaces.
For example, the forward long-range MEMS LIDAR ML-Xs launched by Yijing Technology also uses the MEMS micro-galvanometer + 1550nm fiber laser scheme, which is better than the industry average in many performances such as detection distance, wiring harness, and angular resolution.
Through a number of innovative solutions, Yijing Technology not only avoids the problem of shortening the life of the radar caused by the high-speed rotation of the mechanical lidar motor, but also breaks the traditional structural limitations of MEMS lidar in one fell swoop.
In view of the problem that MEMS lidar is weaker than mechanical lidar in terms of horizontal perception field of view due to the small rotation angle of the micro-galvanometer, yijing technology has specially designed the optical path with the laser reflected by the micro-galvanometer with a special optical angle expansion system.
At the same time, based on the self-developed reception and ASIC chips, and with the signal processing algorithm, Yijing Technology not only realizes key functions such as sunlight immunity and crosstalk light suppression, but also ensures the degree of integration and greatly reduces the cost of lidar.
There is no doubt that traditional mechanical rotary lidar can solve performance problems, but it also brings high maintenance costs (reliability), vehicle-level thresholds, and large-scale manufacturing problems.
MEMS has been widely used in industry, automotive and other industries, but the problem that has been plaguing is that the design of LIDAR based on MEMS corresponds to the mirror size. In order to capture as much light as possible, a large aperture (as large a mirror as possible) is required.
However, the size of the mirror is also limited by certain factors, such as the number of photons received (depending on how many photons are emitted to have a sufficient number of photons returned), collimation, deflection angle (the lens's light deflection), and resonance frequency.
This requires lidar companies to innovate the layout of mirrors and optics, including laser light sources, photodetectors, and lenses, while optimizing the transmit/receive mechanism to achieve ultra-wide horizontal scanning angles.
Obviously, the design of any current lidar scheme will face a real technological innovation threshold before entering mass production. Structural design, breakthroughs in core components, and customization from the application scenario are all crucial.
<h1 class="pgc-h-arrow-right" data-track="94" >2</h1>
In the development path of lidar, another problem that cannot be ignored is that if the implementation does not increase the number of sensors, it can achieve the ability to provide high-resolution long-range detection and the middle-range detection capability of wider field of view (FoV).
For example, in the millimeter wave radar market, Delphi's Multimode electronic scanning radar ESR provides two measurement modes at the same time, integrating a wide viewing angle medium distance and a narrow viewing angle long distance in one.
A company called Sense Photonics (ouster has now signed a definitive acquisition deal with the company) launched the MultiRange concept last year, a FLASH lidar solution based on VCSEL and SPAD.
At the same time, the company proposes key points for the performance evaluation of lidar.
Detection distance is one of the key indicators that have long been used to define the performance of lidar systems. However, from a professional point of view, maximizing performance in one dimension (such as detection distance) comes at the expense of other key metrics (such as resolution, field of view, refresh rate, etc.).
From a physical logic point of view, maximizing the performance of any one metric will have an impact on the performance of the other three metrics. Points per Second (PPS) is a very useful metric proposed by Sense Photonics for quick evaluation of lidar systems.
All along, whether mechanical or electronic, these systems have inherent defects in imaging objects, resulting in blurred motion. Because the laser beam only illuminates a small "point" rather than the entire scene, the laser needs to scan the entire FoV. To overcome this problem, the system needs to run at a higher frame rate, but only at the expense of reducing the detection distance or resolution.
This means that whoever can lead in the PPS indicator has the opportunity to meet the needs of more application scenarios.
For example, AT128, Hesai's long-range hybrid solid-state lidar for ADAS front-loading mass production, is also the only vehicle-grade front-loading mass-produced lidar on the market that meets both long-range (200m@10%) and ultra-high point frequency (1.53 million per second, single echo).
Through the chip-based 128-channel solid-state electronic scanning, the Hesai AT128 realizes the structured scanning of "true 128 lines", avoids the impact of two-dimensional high-speed mechanical scanning on product reliability and life, and realizes the complete field of view of the point cloud in the horizontal and vertical directions without stitching and even distribution, which is similar to the structured data of the camera.
For the high resolution achieved only in a small angle range, or through the dynamic adjustment of ROI to achieve local encryption such as the "indicator trade-off" way, Hesai believes that it can not comprehensively improve the perception of the system, but may bring more uncertainty due to inaccurate ROI selection.
For example, 100 photons go out and only 10 photons return. Only when a vehicle with a reflectivity of 10% can be seen at 250 meters, it is possible to see low-reflex tires on the side of the road at 150 meters, a cardboard box of 20 × 20 cm, and it is possible to see small black (lower reflectivity) objects about 100 meters closer.
In addition, on this basis, Innovusion lidar also has a dynamic focus function, through local pixel encryption, key targets and small objects in the area of interest to "gaze" to obtain more accurate three-dimensional information.
Some industry insiders also admit that as lidar gradually integrates into the pre-existing mass production perception system, it will be the trend to adopt a more comprehensive approach when evaluating lidar systems. "How to make better and faster decisions in the perception system will be the threshold for mass production."