A few days ago, Nissan Automobile's next big Pro-Pilot demonstration vehicle with lidar was publicized in a relatively large range, which is very interesting, unlike the other two Japanese companies, Nissan is currently developing around its own AD/ADAS advanced technology research and development department, which is iterative on the original Pro-Pilot 2.0 by adding Luminar Lidar sensors, combined with nissan's situation, I sorted out the different ideas of the three Japanese companies.
● Nissan
Nissan also assumes a very important development role in the alliance of Renault, Nissan and Mitsubishi, at Qualcomm's autonomous driving chip conference, the RNM alliance cooperated with Waymo before, and did not push down particularly fast, and now with Renault began to sell most of Nissan's equity, from the division of labor, Nissan and Renault in the EE architecture, led by Renault, in the automatic driving development of HPC hardware and software, led by Nissan.
● Toyota
Toyota's current pro-son is Woven Planet Group, with the investment in Aurora and Pony to make a backup, and Toyota's self-research team generally focuses on the development of L2+, in fact, Toyota's cockpit now needs to be understood for a long time, this huge investment in the car soul business may still rely on the American research team more.
●Honda
Honda's Legend development is more like the final struggle of the Japanese autonomous driving team, whether it is in cost control, it is difficult to keep up with the iterative development of the control system; so Honda and GM or full investment in Cruise, but also rely on the capital power of the two to support an independent autonomous driving development company.
Overall, here to see that RNM wants to keep up with the previous few in terms of automatic driver assistance technology, mainly depending on the continuous investment, the current Renault (a loss of assets and sales in Russia) and Nissan from a financial point of view of the blood loss is more serious.
Remarks: For a passive house in Shanghai, I have been at home for the past two months, but I hope that all readers will even jump on this 5-day Labor Day of mine, happy holidays for all readers! Festivals have the right to choose not to stay at home, and to have a good time hanging out!
Figure 1. Three Japanese companies developed a layout for autonomous driving assistance
Part 1 Nissan's automatic assist system developed
Nissan's previous Pro Pilot was based on external vision algorithms, and the entire work was mainly integrated, and this time and the perception configuration to integrate lidar into the new generation of ProPilot is as follows:
(1) 10 cameras
This installation scheme is similar to the usual practice of installing a perception kit in the L4, and is currently a bit like adding a top card to a taxi. 8 cameras in ProPilot 2.0 (3 x front view + 4 AVM + 1 DMS)
1 front-facing camera for field of view detection of a wide range of FOVs in front
2 front-view cameras on the left and right sides of the roof (change position)
2 rearview cameras on the left and right sides of the fenders (new)
1 rearview at the roof antenna (new)
Here, six cameras are used to cover the 360-degree field of view of the whole vehicle, which is 3 more than the original.
And equipped with four round-view fisheye cameras (the number remains the same), front and rear + left and right rearview mirrors (may be integrated with the CMS import)
(2) 7 millimeter-wave radars (2 more than the original)
From the layout point of view, 1 long-distance radar and 4 short-range radar are arranged in 4 corners of the vehicle, adding 2 long-range radars in the rear, and what use of the latter two is really not seen, it may be that the speed information of the rear vehicle is needed when changing lanes
(3) 1 lidar (new)
The lidar uses luminar's lidar and is placed on the roof of the car in front
In addition:
1 x HD map module
Ultrasonic sensors (9): This does not significantly improve functionality
Figure 2. Nissan's perception configuration
The following figure may be seen more clearly, Nissan piled up a circle of surround view cameras, in the fusion algorithm through radar, camera and lidar to do fusion, and called "ground reality perception". At present, Nissan has not said how the entire processor will be processed, and it is still possible to introduce Qualcomm Snapdragon Ride in the future.
Note: Qualcomm and Renault first announced that due to the consistency of the hardware platform, Nissan may have chosen first in this regard
Figure 3. Nissan's algorithm self-research situation
Figure 4. Demonstration of the effect of fusion
In the RPM alliance, there is a unified EE electrical and electronic architecture, and in the 2026 RNM alliance, the next generation of models will use a universal electrical and electronic architecture, if you follow this design, the entire RPM vehicle modeling design will have a huge challenge, I actually did not understand how to solve the above.
Figure 5. RNM unified electrical and electronic architecture
That is to say, in 2026, it will be compatible with centralized electrical architectures.
Figure 6. Centralized electrical and electronic architecture HPC +Zonal+Sensor architecture
This is a cost savings for both sides in the joint development process.
Figure 7. Nissan's main role here is that of Nissan Development
Part 2 Nippon's U.S. strategy on autonomous driving
(1) Toyota
With Toyota's $550 million acquisition of Lyft's self-driving division, Woven Planet's development path is also trying to shift. For Toyota, if many vehicles on the market can collect diversified driving data, it will have great practical significance for the North American team to develop self-driving car systems. Collecting data in Woven-based legacy lidar, with Lyft data buried on operating vehicles (collected in cameras installed), can help improve the high-definition maps used to navigate autonomous vehicles, including intersections, cyclists, pedestrians, and Lyft drivers out of the car. Through these Wovens, a combination of 3D computer vision and machine learning is used to automatically identify traffic objects from the camera's footage.
Figure 8. Woven's self-driving division, which woven bought, also continues to work together
(2) Honda
Legend this matter, the significance of publicity is greater, as Honda began to invest in the overall situation, especially to strengthen the research and development cooperation with GM, the two have actually entered the state of substantive advanced technology alliance.
Figure 9. The world's first L3 production vehicle was given to the 100-unit Honda Legend
Summary: My understanding is that in a sense, Japan and the United States have reached a new generation of technology dependence on smart cars, and the Japanese auto industry may be very dependent on the technology development of the United States except Nissan. In a few years, the Apple series of car software suites will come out, and it may really be possible to change the disk surface quickly.