TEL: +86 23 61663106

A Brief Look at Lifan’s Automatic Parking Matched with Octa-core CPU of ZF


During the past three decades, the revolutionary development of auto industry globally has almost been pushed forward by electric technology, no matter for power system, chassis, air conditioning or lamps. There is an electronic module or “computer ”on the back of these systems which determines the overall performance by its computing power. Today, the electronization is penetrating into auto industry which belonged to the field of mechanical engineering before, and the main engine plants and appliers are now planning on the transformation corresponded to this mainstream tendency while the IT companies then start to connect with auto industry. As the giant of each field, Baidu and ZF developed an automatic parking technology jointly on the first weekend of 2018 by taking advantage of the 330EV from Lifan Motors.


This car--330EV from Lifan Motors has crossed thousands of miles from Chongqing plant to Silicon of United States in order to learn the automatic driving or automatic parking technology precisely from Baidu Institute here,and its goal is to find the space available by itself and has itself parked automatically. Technically speaking, this seems far from those test cars realizing the technology of automatic driving, however, for the limitation of several factors, almost all the test of automatic driving can not be applied to our everyday life. Therefore, all the cars matched with automatic driving including Tesla today need the assistance of driver who is able to monitor the condition of cars and roads.


Why shall this car of Lifan Motors obtain the ability of automatic parking?
Actually, this Lifan 330EV belongs to Pand Auto, which is a company specified at periodical leasing and has its leasing service expanded in 7 cities domestically. By having a rough look at Pand Auto, it seems no difference to those venture-capital companies, while actually, there is a traditional auto corporation--Lifan Motors, so it is reasonable to make periodical leasing business by Lifan cars, and the reason for taking 330EV is due to the operational cost for electricity is much cheaper than that of petroleum for the same miles. Of course, the efficiency for charging is not within our consideration for the time being.


The main function of automatic parking is to solve the problem of car returning of users. During the actual use of periodical leasing, parking is a big trouble for the majority of users, so Pand Auto starts to consider on the method of parking after reaching destination.

Several techniques necessary for the realization of automatic parking on this car?

Being clear about where you are;
Being clear about where you are going;
Being clear about the actual road condition;
Being clear about where the parking space is available;
Being able to have itself parked at the right place;

So, first of all, we’ll go to the first one--Being clear about where you are. This only requires a HD map and accurate GPS positioning, actually, Baidu itself enjoys the qualification and ability of surveying and mapping (another map applier in China --Gaode Map is also actively promoting the accumulation of data), therefore, what the periodical leasing only needs is the finished HD map scanning.

For the destination problem, it is actually on the foundation of route planning of map, whose logic is quite similar to that of GPS--from point A to point B and the car just needs to go as the map pointed.

During the process of parking space searching and car parking, there will definitely be some cars and passengers, so it is a big problem to avoid traffic accidents. In the active safety test we’ve made before, this problem was fully reflected. The mainstream types of sensor are camera, milimeter - wave radar. The actual speed will be within 5km/h, so all we need is just one camera with image analysis for the road condition is relatively simple when we are using automatic parking.


Relatively speaking, it is much easier to find the parking space and have it parked well automatically, because it just needs several hardwares like all the common automatic parking system, what’s different is that we need a few more reversing radar for making accurate judgment on distance and identifying the parking space then. In the meantime, the electric steering can adjust the direction in terms of distance. Most importantly, all the transmission of electric vehicles are automatic, which means we can go forward or back with the controlling of computer. For the type of mechanical gearbox, we need to add a set of electronic controlling system principally speaking, or the automatic parking can not be realized in terms of hardware structure.

All that mentioned above is the premises for the realization of automatic parking , while the key is to connect them all together effectively, for each system shall be under the management of a conjunct brain, no matter the camera, steering system or braking system, actually, the latest CPU of ZF--ParoAI can perfectly satisfy this requirement.


8-core CPU from ZF
Briefly speaking, ZF ProAI is a computer with 8-core CPU, which is made up of over 7 billion semiconductors, as we’ve mentioned before, the semiconductor manufacturing technology of Germany can limit the diameter to 10 naometers at minimum, which is ten times smaller than HIV virus. It can compute 300 thousand times per second. Although it is far from the super computing ability of some other CPU, it is enough for us allowing for the power consumption of only 30w which balances the direct relationship between power and volume.

In my opinion, this cooperation among ZF, Baidu and Pand Auto can not totally represent the power of ProAI, while for the actual power releasing efficiency. Our engineer hasn’t provided the accurate number. There are 6 milimeter-wave radars, double-eye or tri-eye camera, GPS signal with HD map, a set of laser radar, and these sensors update every 40 miliseconds, saying that it has scanned the circumstances for 5 times within just a blink. Therefore, the realization of automatic driving is on the foundation of massive data exchange and computing. The ProAI is an open structure which can carry them all, and this kind of development concept and performance coincide with that of Baidu Apollo, which is an important reason why the bilateral parties can reach the goal so quickly. (Both parties signed strategical agreement in autumn of 2017 and the automatic parking is realized just 4 months later.)