Drivers rely on their brains for cognition, but the “brain” of driverless cars is a computer. The computer in the driverless car is slightly different from the desktop and laptop that we often use, because the car will encounter bumps, vibration, dust and even high temperature when driving, and generally the computer cannot run in these environments for a long time. Therefore, unmanned vehicles generally use the industrial environment of the computer – industrial computer.
The industrial computer runs an operating system that runs driverless software. Figure 1 shows the software system architecture of an autonomous vehicle. On top of the operating system are support modules (here modules are computer programs) that provide basic services to the upper-layer software modules.
The support modules include: Virtual switch modules for inter-module communication; Log management module, used for log recording, retrieval and playback; The process monitoring module is responsible for monitoring the running status of the whole system. If a module runs abnormally, it will prompt the operator and take corresponding measures automatically. The interactive debugging module is responsible for the interaction between developers and the unmanned driving system.
In addition to being aware of the outside world, machines must also be able to learn. Deep learning, which is an efficient machine learning method derived from artificial neural network, is the foundation of the success of driverless technology. Deep learning can improve the time efficiency of vehicle identification of roads, pedestrians, obstacles, etc., and ensure the correct recognition rate. After training with a large amount of data, the car can transform the collected information such as graphs and electromagnetic waves into usable data and realize driverless driving with deep learning algorithm.
When the driverless vehicle collects data through radar, the original training data should be preprocessed first. Calculate the mean and normalize the mean of the data, do principal component analysis of the original data, and use PCA or ZCA bleaching. For example, the time data collected by the laser sensor is converted into the distance between the car and the object; The photo information taken by the car camera is converted into the judgment of the roadblock, the judgment of the traffic light, the judgment of the pedestrian, etc. The data detected by the radar is translated into the distance between the various objects.
The application of deep learning in driverless vehicles mainly includes the following steps:
1,Prepare data, preprocess data and select appropriate data structure to store training data and test tuples;
2,Input a large amount of data to conduct unsupervised learning on the first layer;
3,Through the first layer of data clustering, similar data are divided into the same category for random judgment;
4,Use supervised learning to adjust the threshold of each node in layer 2 to improve the correctness of data input in layer 2;
5,A large amount of data is used to conduct unsupervised learning for each layer of the network, and unsupervised learning is used to train only one layer at a time, and the training results are taken as the input of the higher layer.
6,Use supervised learning to adjust all layers after input.
Application of artificial intelligence in information sharing of autonomous driving
First, wireless networks are used to share information between cars. Using a dedicated lane, a car can share its location and road conditions with other cars in the queue in real time, so that their self-driving systems can make adjustments as they receive the information.
Secondly, it is 3D road condition sensing. The vehicle will combine ultrasonic sensors, cameras, radar and laser ranging technologies to detect the terrain within about 5 meters in front of the car, determine whether the road ahead is asphalt or gravel, grass, beach and other roads, and automatically change the car Settings according to the terrain.
In addition, the car will be able to shift gears automatically, slowing down when it detects changes in the terrain and returning to its original condition when the road returns to normal.
The amount of traffic information collected by car information sharing will be very large. If these data are not processed and utilized effectively, they will be quickly forgotten by information. Therefore, data mining and artificial intelligence should be used to extract effective information and filter out useless information. Considering that the information that the vehicle needs to rely on in the process of driving has great temporal and spatial relevance, some information processing needs to be very timely.
The advantages of artificial intelligence applied to autonomous driving technology
Ai algorithms focus more on learning, while other algorithms focus more on computing. Learning is an important embodiment of intelligence, and learning function is an important feature of artificial intelligence. At present, most artificial intelligence technologies are still in the stage of learning. As mentioned above, unmanned driving is actually human-like driving. It is an intelligent car that learns from human drivers how to perceive the traffic environment, how to make use of the existing knowledge and driving experience to make decisions and planning, and how to skillfully control the steering wheel, accelerator and brake.
From the three aspects of perception, cognition and behavior, the perception part is the most difficult, and artificial intelligence technology is most applied. Sensing technologies that rely on sensors, such as cameras, are popular in the industry because of their low cost. An Israeli company called Mobileye has done a great job of identifying traffic images. It uses a camera to identify traffic lines, traffic lights, pedestrians, and even distinguish between a bike, a car, and a truck.
Deep learning is the most successful application of artificial intelligence technology in the field of image recognition. In recent years, researchers have trained image samples through convolutional neural network and other deep learning models, greatly improving the recognition accuracy. Mobileye’s success so far is due to its early embrace of deep learning as a core technology. In terms of cognition and control, the traditional machine learning technology in the field of artificial intelligence is mainly used to build a driver model by learning the driving behavior of human drivers, so as to learn how to drive the car.
The challenges and prospects of driverless technology
In the context of worsening traffic conditions, the commercial prospects of “driverless” cars are still restricted by many factors.
2,Common agreements are established between models of different brands, and the industry lacks norms and standards
3,Basic road conditions, identification and information accuracy, information network security
4,Unaffordable high costs
In addition, one of the most important characteristics of “driverless” cars is the high level of network and information, which also poses a great challenge to the security of computer systems. In case of computer program disorder or information network intrusion, how to continue to ensure the safety of their own vehicles and other vehicles around, this is also an urgent problem to be solved in the future. Although there are still many challenges in driverless technology, it is difficult to perceive and focus on “learning”. Sooner or later, the level of driverless technology will surpass that of human beings, because stability, accuracy and speed are innate advantages of machines, which cannot be matched by human beings.
Driving is sometimes not a burden, but a pleasure, an expression of the human ability to stretch our limits. The author believes that complete driverless driving may be a little far away, but with the improvement of machine learning algorithms and application mining, a more down-to-earth, human-machine harmonious driving is just around the corner. No matter how difficult it is on the road of autonomous driving, BUT I believe there will always be a day when it will appear on the city road, the development of technology is full of passion and power. In the near future, maybe autonomous driving will become mainstream.