Nissan Motor Co. yesterday officially announced the latest development progress of the new driver assistance ProPilot.
Before we talk about what progress this system has made, let's review the history of proPilot and look forward to its tomorrow by looking at Nissan's autopilot yesterday and today.
In July 2016, Nissan is introducing the ProPilot 1.0 Driver Assistance System, which enables ACC adaptive cruise capability with sensors such as a front-facing monocular camera and millimeter-wave radar, and also has some lane-keeping capability, but the driver must rest his hand on the steering wheel when using this feature.
Three years later, proPilot 1.0 officially came to China, and Nissan's star models Tianlai, Qashqai and Qijun were equipped with this function. But at that time, ProPilot 2.0 had been launched in Japan, and compared to the 1.0 version, Nissan's expectation for ProPilot 2.0 was to face Tesla Autopilot head-on.
Let's take a look at the proPilot 2.0 upgrades, and by convention we first look at the sensor configuration:
5 radars (1 front + 4 angles);
12 ultrasonic radars;
8 cameras (3 x front view + 4 AVM + 1 DMS);
Zenrin high-precision maps.
Relying on this sensor configuration, Nissan has implemented a high-speed navigation assisted driving function and a single-lane off-hand driving function. These two functions may not be very strange today, and the new domestic force Wei Xiaoli has delivered high-speed navigation assisted driving. As for off-the-hand driving in a single lane, we are still cautious in this regard due to the different traffic conditions in different places, but from a technical point of view, it is not difficult to achieve.
Then, for Nissan in 2019, the delivery of navigation assisted driving and single-lane off-the-hand driving is undoubtedly the best embodiment of advanced technology research and development. Looking at the world, in addition to Tesla, no car company has been able to do it.
And 2019 is a delicate time, when the discussion about whether lidar can get on the car has already begun.
Tetsuya Iijima, then general manager of advanced technology development at Nissan ADAS, said in an interview:
Lidar lacks performance that surpasses the latest technology in radar and cameras, and it would be great if lidar could reach the level of application in automotive systems. But this is not the case, and its cost and performance do not match.
The information we can get from Tetsuya Iijima's words is that Nissan is not exclusive of lidar, and if one day in the future, the performance of lidar is improved, and the cost can be as low as mass production in large quantities, then Nissan will still consider it.
The day of "lidar cost reduction and performance improvement" did not make us wait too long, and Nissan also launched proPilot 3.0 in a timely manner.
01
Lidar is coming
Compared to ProPilot 2.0, a significant change in Nissan's latest generation of perception architecture is the addition of a lidar. Its perceived hardware is as follows:
10 cameras;
5 mmWave radars (1 forward + 4 angles);
1 x LiDAR (luminar supply);
HD map.
It is worth noting that the camera position of Nissan ProPilot 3.0 is relatively special, and its side front view camera position is similar to that of ET7, placed in the two corners of the roof, showing a "horn" shape. In addition, there are two cameras facing forward and backward at the rearview mirrors of the side fenders of the test vehicle, which indicates that ProPilot has overlapping and redundant perception of the side front.
The LiDAR supplier of Nissan ProPilot is Luminar, but Nissan says it uses the next generation of high-performance lidar, which indirectly indicates that it is not Iris that is being used at this stage. In terms of parameters, Toru Iijima, director of nissan ADAS' advanced technology development department, also said that they believe that the maximum perceptual distance of lidar needs to be more than 300 meters to play a role on the highway, and the horizontal FOV angle is 120 degrees, the vertical viewing angle is greater than 25 degrees, and the resolution is 0.05 degrees.
Compared with Luminar Iris, the lidar used by Nissan is mainstream, except that it is not dominant in terms of detection distance. Here we can also compare this lidar with the lidar used by the new domestic forces.
In addition, Toru Iijima, director of nissan ADAS' advanced technology development department, also talked about the location of the lidar installation, and recently two domestic CEOs li xiang and Xia Yiping also commented on this issue and exchanged views. Iijima also believes that the layout of lidar mounted on the roof, that is, the watchtower style, is the most reasonable, because "the height of the station can see far" and the laser beam will not be blocked.
So what can the system's L4/L5 architecture do? According to Toruya Iijima, director of Nissan ADAS' Advanced Technology Development Department, after adding lidar, this ground truth perception road reality perception can monitor road conditions in real time and can quickly respond after identifying travel obstacles on the vehicle's established driving route.
In the demo video, we were also able to see how the test vehicle equipped with the ProPilot system performed in a real-life scenario.
Continuous "Ghost Probe"
During the driving of the test vehicle, the vehicle in the roadside parking space suddenly reversed, giving the test vehicle a "interception", and the test vehicle quickly slowed down and actively avoided to the next lane. However, there are pedestrians passing through the flower beds in the next lane, at which point AEB triggers and the vehicle brakes urgently.
Maybe you will feel a little strange after watching this scene, because it is not in line with human driving habits, and we have always advocated "giving way to speed". But don't forget, this is an "autonomous driving" model with 360-degree perception.
Avoid road movement obstacles
Another scene is a tire suddenly rolling in front of the vehicle's driving route, which is insoluble at this stage, but after the addition of lidar, the size, shape, and distance of foreign objects in the road environment can be clearly perceived, so the vehicle can make decisions based on information. In addition, stationary obstacles on the road can also be identified and actively avoided.
Iijima also said that the system will still use high-precision maps, and with the blessing of lidar, even if some of the road map is lost, the system can continue to work, and robustness is guaranteed.
Vehicles in navigation assistance can also make active decisions based on the actual situation of the road, and Nissan's new generation of road reality perception system seems to be an enhanced version of the current stage of navigation assistance driving, perfectly covering the "sudden scene".
But in the end, Nissan said that this road reality perception will not be developed until the mid-2020s, that is, around 2025, and will be assembled on all Nissan models in 2030. This means that consumers will have to wait at least 8 years for the system to be used. In an instant, my mood changed from "anticipation" to "discouragement.".
02
The three brothers went hand in hand
The entire Japanese auto industry is less sensitive to the shift from global cars to electrification and intelligence. It is a fact that it lags behind other regions in terms of transformation. But the black words are good, not afraid of "messing up", afraid of "worse than rotten". When it comes to Nissan's research and development results, we have to compare the efforts of two brothers.
But before we introduce the performance of the other two, let's look at what efforts Japanese politics have made to develop autonomous driving.
In 2014, the Cabinet Office of Japan, led by the Ministry of Internal Affairs and Communications, the Ministry of Economy, Trade and Industry, the Ministry of Land, Infrastructure, Transport and Tourism, the National Police Agency, Toyota, Honda Institute of Technology, Nissan and other OEMs, universities and suppliers, announced the establishment of an integrated project between government, industry, education and research to promote autonomous driving projects on a national scale.
In September 2018, Japan's Ministry of Land, Infrastructure, Transport and Tourism established safety technical guidelines for autonomous vehicles;
In March 2019, in order to accelerate the commercialization of autonomous driving and intelligent transportation for the upcoming Tokyo Olympic Games, the Japanese Cabinet passed the amendment to the Road Transport Vehicle Law;
In April 2020, Japan began implementing the Corrected Roads Act, which includes autonomous driving content.
Under the full impetus of this political research and development, Japan's first L3 level autonomous vehicle was officially launched, which is the Honda Legend.
Nominal L3
Before Legend, Audi had announced that its model A8 was the world's first model with L3 level autonomous driving capabilities, but due to regulatory issues, the "first" status of the Audi A8 was not recognized. The Honda Legend is certified by the government and has a proper name, and even the rear of the car is labeled exclusively.
So, the Honda Legend is really "out of hand", what about the configuration and price? This has to take a look at the sensor configuration of this perception system:
Front-facing binocular camera;
5 lidars (front 2 rear 3);
5 mmWave radars;
Rely on high-precision maps;
Saray Electronics R-Car V3U chip, up to 60 TOPS.
Yes, you read that right, this car is equipped with 5 lidars, the model is Valeo's second-generation SCALA, this lidar has only 16 lines, a maximum detection distance of 150 meters, and a horizontal angle FOV of 10 degrees. In addition, according to the installation location, it can only play a role in blindness and redundancy with other sensors.
Most fatally, there are two prerequisites for using Legend Assisted Driving, namely the use of road sections that require coverage of a high-precision map and a speed limit of less than 50 km/h. The first condition limits the scope, the second condition limits the scene, and the so-called L3 is instantly eclipsed and reduced to chicken ribs.
However, Honda originally did not expect to mass-produce the Legend, this car is limited to 100 units, users can not buy out, can only be discounted at a price of 660,000 yuan to lease and use, the lease period is 3 years. It can be seen that the emergence of Legend is the "self-consolation" of Japan's autonomous driving industry.
Throw money at the same time, and develop yourself
Did Honda make an L3? Not only has it been done, but the government has also recognized it. However, from the perspective of available value, the Honda L3 does not have a high gold content. What about Toyota?
When I looked at Toyota's autonomous driving layout, I did find that Toyota's development path was very different from Honda's. Toyota is spending money on the L4 Robotaxi while increasing the popularity of the L2.
Here's toyota's moves in the L4 industry (some of which have been made public):
In 2018, Toyota established tri-AD, an autonomous driving R&D team, which focuses on autonomous driving software research and development;
In August 2018, Toyota invested $500 million in Uber; but Uber's project stalled due to an accident, and in 2020, Aurora acquired Uber's autonomous driving business, and Toyota's cooperation with Uber continued;
In February 2020, Toyota invested $400 million in domestic autonomous driving start-upSma Zhixing;
In July 2020, Toyota's self-driving team TRI-AD announced the establishment of an independent company, Woven Planet, not long ago, Woven Planet's company Woven Alpha had the idea of following Tesla's adoption of a pure visual solution in the direction of autonomous driving, and the team has begun to develop a low-cost self-driving solution.
In 2021, Woven Planet participated in the Series C financing of autonomous driving company Nuro, the specific amount of which was not disclosed;
In April 2021, Toyota acquired Lyft's autonomous driving division for $550 million.
In the research and development of autonomous driving, Toyota has chosen a strategy of self-development and investment and acquisition in parallel, eggs are not placed in one basket, and "casting a wide net" improves the probability of success.
But the L4 is still high, and Toyota is also trying to increase the popularity of the L2. If you look at the configuration table of Toyota models, even low-end models will be equipped with TSS 2.0 system, which can provide drivers with lane keeping, adaptive cruise and other functions. Although in terms of experience, this system cannot be compared with the ADAS system developed by the new forces, but Toyota's use of actions alone to increase the popularity of L2 assisted driving and active safety configurations is commendable.
As for more advanced assisted driving, Toyota is not unable to get on the car, but in the short term, it is limited by factors such as cost, and the average user cannot experience it.
Toyota provides a driver assistance system called Teammate for the Lexus LS 500h and Mirai top-mount users in overseas markets, which has the functions of light lane change and navigation assistance, and is also an L2 system in terms of functions and capabilities.
Through comparison, you will find that the three Japanese brothers are going hand in hand on the road of automatic driving, but Nissan's 2030 loading car will still make people take a breath of cool air.
Domestic new forces do not like to judge ability by SEA standards, but scene landing is the most able to reflect the real ability. The three strongest points of this year's new forces are all urban navigation auxiliary driving, with a lower takeover rate to reflect stronger auxiliary driving capabilities. This is followed by high-level assisted driving through point-to-point.
Judging from the intensity and rhythm of the new forces' "inner volume", Nissan's 2025 research and development landing and 2030 loading on the road seem to have already announced its failure in China before it began. By the way, the above time point refers to the rhythm of R&D and mass production in Japan.
03
Write at the end
I have always believed that assisted driving should not pursue the "bright light" under special conditions, but should strive to pursue the "silent moisturizing" in daily driving. On the road of autonomous driving, Honda "came out on top" and took the lead in mass production of L3, but the system has gradually deviated from the essence of "assisting user driving".
Based on the L2, Toyota continues to attack the L4, impeccable in its development path, but for a long time it has not been able to bring users low-cost and good experience of assisted driving.
Nissan's new architecture enables navigational assisted driving while giving the system more AEB capability, but it is still on the testing ground. Judging from the time strategy of 2025 and 2030, if you want to gain a place in China, the main battlefield of auxiliary driving, you still need to make all-out efforts.
Author: Bobo
Edit: June 3
These are also worth reading