Where is the lidar installed? This was originally a topic that was not a topic, but it was detonated by the liftable "jumping light" dual lidar scheme announced by Jidu.
(Jidu's first "automotive robot" concept front cover dual lidar)
First Li Xiang and Xia Yiping shouted in the air, and then He Xiaopeng joined the polemic to support Xia Yiping, and now there is a founder of a new car-making force to comment, rubbing a wave of heat.
Under the trend of escalating environmental perception technology, it is not only the reliance on lidar that is the only way. In the entanglement of "loading 1 or 2", "loading the roof or loading the head", in fact, the set has realized the "lidar + pure vision" automatic driving program, walking in the forefront of the industry.
It seems that the "immortals fight" like shouting in the air, but in fact, the "immortals" are not in a Taoist line at all.
Where the lidar is installed depends on the functional realization and the beauty of the shape
At present, there is no conventional standard in the entire automotive industry, requiring lidar to be installed in a fixed position. Each car company needs to seek a balance between multiple dimensions such as cost and performance, and finally demonstrate a solution suitable for its own autonomous driving function.
Industry insiders also basically believe that the location of lidar installation depends on the needs of car companies, but different layout schemes will have different impacts on the realization of automatic driving functions.
An industry expert told the media that the lidar deployed on the roof of the car can indeed get a better view, but in terms of FOV, the two radars still have an irreplaceable advantage over a single one. "If there is only one 120-degree FOV lidar, we think that it is impossible to complete the intelligent pilot function in the urban scene, and can only complete the intelligent pilot function in the high-speed scene."
(SimuCar 2.0 software-integrated analog prototype with dual lidar installed)
From the perspective of the design of the "auto robot" by Jidu official, it is to achieve high-level automatic driving, and it is "three-domain integration", that is, "high-speed domain, urban domain, parking domain" integration. The set degree adopts the "jumping light" type lidar scheme, which can achieve a horizontal 180-degree FOV coverage, which improves the recognition ability of the "ghost probe" and the left and right obstacles. Since the middle 60 degrees is the overlapping viewing angle of the two lidars, it has a double point cloud for more accurate identification.
(Set degree dual lidar point cloud)
However, it is not yet known where the set will be installed in the lidar, the current exposure is only the concept design, how the real car is still full of suspense.
In addition, many car companies are more from the perspective of engineering, ignoring the strict requirements of users for the United States. In fact, users have a very high pursuit of the shape of the whole vehicle, and even some netizens criticized the "horn" shape of a new force in the car, saying, "If it is too ugly, don't talk."
LiDAR + pure vision achieves "true redundancy" and is safer
The core role of lidar is to solve the problem of environmental perception. Domestic car companies basically integrate the combination of lidar + camera + millimeter wave radar to meet the perception of the surrounding physical world by intelligent cars.
But solving environmental perception is not just a path for lidar. Tesla, for example, does not use lidar, but only uses the pure visual solution of the camera to do automatic driving. Another example is Mobileye, which adopts the "lidar + pure vision" automatic driving scheme, these two are independent subsystems, one uses a pure camera, and the other uses a combination of lidar and radar.
Jidu also adopts pure vision and lidar automatic driving solutions. Officially, it is a separate dual system that can achieve "true redundancy". Compared with the traditional solution, the two sets of automatic driving solutions of Jidu are mutual backup and complement each other, and the two sets of solutions can not only drive themselves efficiently, but also efficiently coordinate. From a security point of view, the set of solutions can better cope with complex scenarios and have better security.
However, given the limited information currently disclosed, it is not enough to prove that the set degree and Mobileye's scheme must be similar, and the underlying logic of the two sides may not be the same.
One thing is for sure, pure visual autonomous driving solutions require sufficient data accumulation and algorithm iteration. It is possible that Jidu has applied the atomization capability of Baidu Apollo's pure visual solution, but it has not yet been confirmed by sufficient strong support.
(Baidu Apollo Robotaxi)
The rich accumulation is certain, and Baidu also has nearly 10 years of technical accumulation in the field of automatic driving. Apollo's Robotaxi has L4 fully unmanned driving capabilities, has 25 million kilometers of safety 0 accidents of automatic driving test mileage, a large number of Robotaxi fleet in more than 30 cities across the country to continuously carry out real road tests, based on massive data and constantly iterate algorithms.
From the perspective of technological advancement, the lidar and pure visual automatic driving solutions of the concentration are not only safer, but also have rich technology accumulation and long-term test verification, and are in a leading position in the industry.
Instead of dwelling on where the lidar is installed, or sending microblogs to rub the heat, in fact, there is no sense, it is better to increase investment in research and development, roll up the sleeves and work hard, create more subversive technical solutions, and achieve industry leadership.