Wen Hao Ott is driving? Yang Ming
Chihiro Positioning The world's largest high-precision positioning road test has achieved initial results.
Recently, Chihiro Location official news said that the road test team has covered nearly half of China, including Beijing, Shanghai, Han, Hangzhou, Chongqing, Kunming, Chengdu, Changsha, Guangzhou, Xi 'an and Lanzhou, as well as G 15, G40, G42 and G50, with a cumulative mileage of over 90,000 kilometers.
* Chihiro position in national road test.
This large-scale road test effectively verified the stability of high-precision positioning service in different latitudes, different altitudes, different road conditions, different weather and other scenarios, and provided massive data support for intelligent driving algorithm optimization iteration.
The results of road test show that Chihiro can locate the specific lanes of vehicles driving in Hangzhou and provide lane-level navigation information under the coverage of high-precision maps.
* Chihiro's location road test in Hangzhou.
In Guangzhou, Chihiro's high-precision positioning results are still accurate, but due to the serious ionospheric interference, the satellite positioning error is large.
* Chihiro took a road test in Guangzhou.
In Chongqing, even though there are many forks, viaducts and tunnels, Chihiro positioning can still provide accurate, safe and reliable positioning results. Especially in the case of long-term star loss in the tunnel, the combination algorithm of Chihiro positioning self-developed satellite and inertial navigation can still ensure high-precision positioning effect, making it possible to drive automatically in the whole scene.
* Chihiro's road test in Chongqing
Nian Jin Fei, general manager of Chihiro Location Intelligent Driving Division, said, "Large-scale road test is long-term and continuous. Through large-scale road tests, high-precision positioning can provide very high-precision position information output. This positioning ability is just needed for automatic driving above L3 level, which has played a blessing role for L2 level assisted driving. "
Chihiro positioning how to optimize the positioning algorithm and improve the positioning accuracy of the car to 20 cm?
Link: "Algorithm, the secret of more accurate positioning, here? →? The fourth station internal test "
The road test is carried out nationwide, and the positioning accuracy can reach up to 2 cm.
Chihiro's first high-precision positioning large-scale road test in the industry has been verified in all highways and major cities in China, which indicates that high-precision positioning mass production technology will land on a large scale in the field of intelligent driving, providing space-time intelligent infrastructure guarantee for intelligent driving.
During this period, Chihiro's large-scale road test has been carried out for more than half. The test vehicle is equipped with Chihiro's self-developed spatio-temporal intelligent algorithm and high-precision positioning service, and runs all over the national highways and highways in major cities, testing for different weather, different altitudes and multiple scenarios, and constantly verifying and iterating algorithms and services.
Previously, Chihiro had completed a high-precision positioning road test totaling 300,000 kilometers.
According to the current test results, the real-time dynamic positioning accuracy of the test vehicle can reach up to 2 cm, which can well meet the extensive needs of intelligent driving and vehicle-road coordination above L3 level and realize lane-level navigation. The so-called lane-level navigation means that the vehicle not only knows which road it is driving, but also can distinguish which lane it is driving.
So how does Chihiro achieve high-precision positioning and large-scale road test?
First of all, in terms of infrastructure, based on the Beidou satellite navigation system, Chihiro has built 2,800 ground reinforcement stations throughout the country, forming a "national network" of ground reinforcement systems.
In July this year, Chihiro announced that the "National Network" was upgraded again. On the basis of covering most parts of the country before, the national expressway network is fully covered, including the newly opened G7 expressway connecting Beijing and Xinjiang, paving the way for intelligent driving.
For the expressway network under planning or construction, Chihiro will continue to keep up with the pace of national road network construction and simultaneously realize the coverage of Beidou foundation reinforcement station.
It is worth mentioning that the redundant backup between stations is fully considered in the design and site selection, so that when there is a problem in the station, the nearby stations can make redundant backup, and the peripheral service capacity will not decline.
At the same time, the station is built in the form of dual antennas and dual receivers, and a fully redundant transmission link is guaranteed on the link where each reference station returns data to the data center.
Next, let the test vehicles equipped with Chihiro location professional service and intelligent algorithm run on all highways and urban highways in China, and obtain data from actual large-scale tests, thus iteratively upgrading the algorithm.
* Chihiro's technicians install road test equipment.
Chihiro's positioning technician revealed that "the camera and radar in the vehicle may be inaccurate in the case of bad weather or lane line fouling, and there will be errors, which will lead to certain security risks. At this time, if we can combine absolute positioning, that is, satellite positioning, the driving vehicles are not afraid of being misled. In professional terms, it is to combine relative positioning with satellite positioning. "
After the road test data collection is completed, Chihiro will first store a large number of road test data in the cloud, and then the server in the cloud will automatically classify the data, including satellite positioning data, differential correction data, vehicle dynamic sensing data, high-precision benchmark data and so on.
After storage and classification, these data will be transferred to Chihiro location algorithm library, which is equivalent to a big data test field running 7*24 hours. There are many complicated mathematical models in it. Match all kinds of road test data in the algorithm library, do a lot of calculations, and finally adjust and solve the best mathematical model to achieve high precision positioning. Through continuous road test and iterative test of the algorithm, the positioning accuracy is getting higher and higher.
Add a typical scene database for high-precision positioning without dead angle.
Taking the actual test of Yun Qi Town in Hangzhou as an example, in the video released by Chihiro for the first time, we can see that the vehicle position can be clearly seen through Chihiro's high-precision map positioning ability, and the technicians demonstrated the lane-level high-precision positioning ability through the tablet computer.
* Hangzhou Chihiro positioning is based on high-precision maps for high-precision positioning trajectory comparison.
In the demonstration, the blue line is a high-precision positioning trajectory, and the red line is a common positioning trajectory, in which the blue line is consistent with the real-time position of the vehicle. When the vehicle changes lanes, the blue line will change lanes, and the red line has deviated from the road.
* Satellite signal-to-noise ratio of Chihiro monitoring in urban canyon scene test.
At the same time, the signal-to-noise ratio of the current visible satellite signal can be observed in real time through Chihiro's self-developed software installed on the mobile phone in the test car. When the satellite signal is lost or blocked, the positioning error increases. Chihiro positioning uses close combination, multi-constellation, multi-band and multi-sensor information fusion to solve this problem well and improve the availability and reliability of positioning.
Among them, the close combination of Chihiro position is helpful to accelerate the convergence speed, so as to realize the overall improvement of fixed speed, convergence time, accuracy and integrity.
* Guangzhou Chihiro position high-precision positioning trajectory comparison
In different latitudes and climatic environments, such as Guangzhou, the ionosphere is very active in low latitudes near the equator, which will affect satellite positioning and cause problems such as inaccurate positioning. Chihiro position test vehicle is equipped with optimization algorithm to achieve accurate positioning.
In addition, Chihiro also established a database containing dozens of typical scenes, such as tunnels, ramps, toll stations, low overhead bunkers and so on. The mileage is over one million kilometers. Combined with the automatic scene perception algorithm of the terminal, it helps the vehicle to further improve the positioning accuracy.
* Chihiro is located in Chongqing Huangjuewan Interchange Road Test Site.
Tunnel is the difficulty of satellite positioning, because the satellite signal is blocked and the satellite will lose its lock for a long time. Chihiro uses a self-developed satellite integrated inertial navigation system, which is an optimization algorithm specially developed for complex terrain.
* Chihiro is located in Chongqing Huayan Tunnel Highway Test Site.
Simply put, integrated inertial navigation is a fusion scheme of lidar and camera, and it is a relative positioning technology to assist satellite positioning.
When the satellite loses its lock, the program will immediately call the gyroscope in the vehicle, and calculate the position of the vehicle in the tunnel through the collected speed and steering data, so as to solve the blind spot problem in the tunnel, and still ensure the signal integrity and functional safety under the condition of insufficient available satellites.
It can be seen that after Chihiro position high-precision positioning and large-scale road test is completed, self-driving cars can be empowered in many aspects, such as improving the stability and reliability of the sensing system.
At present, the mainstream autopilot sensors, such as cameras, millimeter-wave radars, laser radars, etc., all have relative positioning capabilities, but because they all have their own defects in the actual application process, their biggest hidden danger is the lack of comprehensive information about the scene in front of autonomous vehicles. They only rely on accurate sensor perception and powerful calculation to deal with the problems encountered in each scene, but lack the framework to deal with advanced abstract problems in complex scenes and extreme scenes.
For example, without the aid of high-precision positioning, this sensor scheme will affect the vehicle entering/leaving the adjacent expressway in advance or delay, or the vehicle will swing from side to side when leaving the expressway due to the lack of accurate positioning. This will greatly increase the accident rate of vehicles and also affect the comfort of vehicles.
The performance boundary of the sensor makes some information in the scene imperceptible. The lack of curvature and slope angle in the road will have a great impact on the longitudinal and lateral planning and control of the vehicle, and passengers will feel that the control of the vehicle is not as smooth as that of the driver.
In contrast, if Chihiro positioning technology is integrated, the interference of external factors can be avoided. Even in bad weather, non-line-of-sight scenes, and weak signals from other on-board sensor satellites, it can still provide accurate, stable and reliable positioning information for self-driving cars, and help vehicles to judge their current positions, so as to have an overall understanding of the current environment and facilitate vehicles to make the next decision.
IMT-2020？ (5G)？ In the "White Paper on High-precision Vehicle Positioning" issued by the Working Group of Propulsion Group Cellular Vehicle Federation (C-V2X), it is pointed out that high-precision vehicle positioning plays a decisive role in autonomous driving under the conditions of bad weather, repeated scenes, non-line-of-sight scenes and unstable vehicle sensors.
Nian Jin Fei said: "High-precision absolute positioning based on GNSS is one of the core basic capabilities of smart cars, because smart cars can make correct judgments only if they know where it is, especially its precise location, and then control the next move of the vehicles. At the same time, high-precision positioning can also play the role of security redundancy. Although sensors such as cameras and radars also have relative positioning capabilities, they will be affected by weather and terrain. High-precision positioning can help cars achieve accurate positioning of the whole scene and provide a reliable basis for intelligent decision-making. "
Moreover, through the fusion application of high-precision positioning, camera and high-precision radar, good perceptual redundancy can be realized. For example, Guangzhou Automobile New Energy Aian V is equipped with high-precision positioning services. Relying on the high-precision space-time intelligent service provided by Chihiro, the positioning accuracy of Guangzhou Automobile New Energy Ai 'an V reaches centimeter level, which makes the relative positioning scheme still fail when the vehicle encounters irregular lane lines, short-term coverage of lane lines, no obvious signs on the road, excessive curvature of the curve, etc.
In addition, with Chihiro's world's largest high-precision space-time service network, Xpeng Motors' super-long life intelligent coupe P7 has strong space-time perception ability, and the global positioning accuracy of the whole vehicle can reach centimeter level. On the expressway covered by high-precision map, NGP high-speed autonomous navigation and driving can be realized.
A few days ago, the General Office of the State Council issued the New Energy Automobile Industry Development Plan (202 1-2035), clearly proposing to promote the application of satellite navigation systems such as Beidou in the field of high-precision positioning. "Planning" mentioned in the implementation of intelligent network science and technology innovation project that it is necessary to break through core technologies and products such as high-precision mapping and positioning. With the continuous improvement of the development level of autonomous driving, more and more manufacturers also realize that it is impossible to continuously strengthen the vehicle's ability to deal with complex road conditions and massive data only by the solution of "sensor+control system".
At this time node, Chihiro's first large-scale road test has just ended. At this time, Chihiro's high-precision positioning will become an important part of L3 autonomous driving technology. It can be said that the high-precision positioning on the market has become an indispensable part for many automobile manufacturers to launch high-end autonomous vehicles.
It can be seen that high-precision positioning and deep integration of body sensors will play an increasingly important role. If we compare the evolution of key factors of intelligent transportation, we can find that centimeter-level positioning is the key to realize automatic driving.
This article comes from car home, the author of the car manufacturer, and does not represent car home's position.