Apollo's hard power
It has been about three years since Baidu Apollo came out (20 17), during which the development speed of Baidu Apollo can be described as rapid. In March 2020, by the well-known research institute NR(Navigant? In the ranking of self-driving competitiveness in 2020 released by Research, the ranking of Baidu Apollo jumped to the ranks of international self-driving "leaders", ranking fourth in the world, and the top three were Waymo, Cruise and Ford. But different from the top three, Baidu Apollo has been the challenger of the third camp since 20 17, and also the competitor of the second camp. It can be described as "leapfrog" every year after surpassing many veteran players.
In the 20 19 self-driving vehicle takeover report released by DMV of California Motor Vehicle Administration, Baidu Apollo's test mileage in California reached 108300 miles (about 174292 kilometers), taking over the data MPI (miles? per？ Intervention, the average mileage between every two manual interventions is 18050 miles (29049 kilometers), and the cumulative test mileage and MPI have also increased rapidly, especially the MPI score surpassed Google Waymo for the first time, occupying the top spot.
In China, Baidu Apollo has always been an industry leader. From the beginning, it wants to build not only an autopilot system, but an autopilot ecosystem that can make full use of its own advantages, so that the autopilot system can be established through a perfect test system. Baidu Apollo is responsible for providing and integrating the autopilot technology and resources required by car companies, such as open source code, AI algorithm technology, high-precision maps and a simulation platform for autopilot testing. Car companies can directly use the advantages of Apollo platform to continuously improve and perfect autopilot technology, and finally achieve a win-win situation.
Baidu Apollo platform has reached cooperation with 19 car companies in 20 18, and Robotaxi, a self-driving taxi equipped with Baidu V2X technology, started operation in 20 19. At present, Apollo has formed three open platform systems, namely, autonomous driving, vehicle-road coordination and intelligent vehicle association. It has 36,000 developers worldwide, 178 eco-partners, 560,000 lines of open source code, 1237 smart driving patents and 150 driving test licenses, and has carried out road tests in 23 cities around the world.
Apollo? A footman? What's so special about parking?
In April this year, Baidu Apollo won "new infrastructure" projects in three cities with its own hard power. I have to sigh that even if Baidu is ridiculed by the whole network, the strength of Apollo must be affirmed. So what are the concerns of this Apollo mass production product ACU platform?
ACU computing platform was released by Baidu at the 20 18 GTC conference. According to the demand for computing power in different scenarios, * * is divided into three series of products, including ACU- Basic and ACU- Advanced? ACU-Professional can support 5 cameras, 12 ultrasonic radars, and reserve the interface between millimeter-wave radar and laser radar. Among them, ACU-Advanced is a special vehicle computing platform for autonomous parking product AVP and the first product of Baidu ACU series.
As a dedicated computing platform for Baidu Apollo autopilot, ACU has a very comprehensive AI capability, covering high-precision mapping and positioning, environmental awareness, planning and control and other functions. At the same time, ACU also has cloud update capability, which can realize real-time closed-loop iteration of data return, cloud training and upgrade algorithm. The landing of models equipped with ACU computing platform can realize Baidu AVP autonomous parking scheme, and can also solve the time cost and operating cost that a lot of parking is difficult to occupy.
In this regard, Baidu mainly presupposes two major scenarios. One is Homa-AVP for the fixed scene of family/company: drivers can get off at the elevator entrance of the parking lot directly, and vehicles can park in a fixed parking space along a fixed route within sight; Another scene is to create a public AVP for public scenes such as transportation hubs and commercial centers: drivers can get off at the destination directly, and vehicles automatically enter the parking lot to find parking spaces. If the parking lot is congested, vehicles can also queue up automatically. In addition, equipped with this product, users can also remotely control the car with one button on the mobile phone to realize remote car picking, automatic parking, automatic parking and so on.
From this point of view, ACU Advanced is a domain controller specially responsible for autonomous parking, which can uniformly deploy major sensors and algorithms and focus on autonomous parking. How many parts are there in the autonomous parking scheme of traditional car companies? ECU？ Always go your own way, ACU? It is a highly integrated part and can support high-speed automatic driving as a part of redundancy.
As can be seen from the above two application scenarios, Baidu's automatic parking function is much more powerful than most car companies, especially the second scenario. Why the majority? Because two years ago, Bosch and Daimler demonstrated the fully automatic parking service technology (Automated? A footman? Parking), much like Baidu's version.
Bosch-Daimler's parking service system needs to pre-install sensors and computing equipment in the parking lot. The sensor can sense the environment around the parking lot and the driving path of the vehicle, and then the computing equipment in the parking lot can calculate the driving path of the vehicle and issue driving operation instructions, which will be sent to the vehicle through the network such as WiFi, and the vehicle's by-wire control system can complete automatic parking.
Bosch's sensors are responsible for detecting the driving path of the vehicle and the surrounding environment of the parking lot, and providing relevant information necessary for commanding the vehicle. The cooperation between the intelligent infrastructure of parking lot provided by Bosch Internet technology and Mercedes-Benz automobile technology can transform the instructions issued by the intelligent infrastructure into driving operation instructions. In this way, the car can go up and down the ramp by itself and move between different areas of the parking lot. Once the infrastructure sensor detects an obstacle, the vehicle will also brake immediately.
The owner drives the car to the parking lot, and after getting off the bus, he can drive the car to the parking space by tapping on the screen of the mobile phone. When the owner leaves the parking lot, the vehicle will automatically drive to the designated parking space. When the bus comes back from the event, turn on the mobile phone again and the vehicle will return to the place where the owner got off. The whole process does not require the driver to get on the bus at all. So for most users, Apollo? A footman? Parking is worth looking forward to.
At the same time, according to the relevant person in charge of Baidu, there are currently 2.7? 1 100 million passenger cars only match about 654.38 billion parking spaces, with a gap of about 50 million parking spaces, especially in first-tier cities, where parking spaces are unevenly distributed and the average effective utilization rate of parking lots is only 45%. In addition, in the process of driving, the problem of "difficult parking" is more common. Even for "old drivers", on average, more than 30% of driving time is wasted in the process of finding parking spaces and parking, which reflects a huge market. Baidu claims that this set of self-service parking products can save 70% of the time for picking up and returning cars and get more than 99% of parking spaces, which can completely solve the parking problem.
For a long time, the self-parking function has been somewhat embarrassing for most users, and it is far less skilled than the "old driver" reversing into the warehouse himself. However, with the continuous development of autonomous driving technology, the function of automatic parking has been greatly expanded and improved, and even "looking for a car" and "looking for someone by car" can be realized. These are the points that are easy to attract and impress consumers in real life. As for the application scenarios of automatic parking in the future, let's wait and see.
Figure? |? From the network
This article comes from car home, the author of the car manufacturer, and does not represent car home's position.