[Guide] When we talk about autonomous driving, what role will artificial intelligence play in it? With the rapid development of modern high-tech, digitization, informatization and intelligence are increasingly being applied to all aspects of the production and life of human society. Intelligence that was once only seen in popular science novels
When we talk about autonomous driving, what role will artificial intelligence play in it?
With the rapid development of modern high-tech, digitization, informatization and intelligence are increasingly applied to all aspects of the production and life of human society. The intelligent driverless cars that were once only seen in popular science novels are no longer It is illusory. People will be able to see smart driverless cars in reality in the near future. Nowadays, the performance, comfort and safety of automobiles that integrate various high and new technologies have made great progress. Intelligent driverless cars usually have a highly intelligent computer, which can receive various information about the surrounding environment and the car itself from various smart sensors, and can efficiently and quickly comprehensively organize it, and then transfer the information to the car’s execution system, So as to realize automatic driving, intelligent control and other functions.
The technical principle of auto-driving vehicles
When a human is driving a car, point out the destination on the electronic map of the on-board display screen, design the driving route, and automatically import the completed electronic map confirmation into the central processing unit, and the central processing unit plans a reasonable driving path according to the boundary of the plot.
In the system, angle sensors, motor speed sensors, position sensors, pressure sensors and other sensors are used to measure driving information, and the driving information is converted into electrical signals and transmitted to the central processing unit for calculation, and then the central processing unit issues instructions to control the automatic transmission steering mechanism, forming a closed-loop system to realize intelligent control of car driving.
Artificial intelligence technology in unmanned driving
Behavior decision system or driving decision system, including global route planning and navigation and local obstacle avoidance and danger, as well as conventional traffic rules-based driving strategy (the simplest, keep the car in the lane), the technology used is divided into three categories
1) Technology based on reasoning logic and rules
The A* and D* algorithms of global route planning and navigation, the dwa algorithm of local obstacle avoidance, the conventional optimal control mathematical methods (such as multi-objective decision-making), and the FSM rule engine based on traffic rules belong to this type of technology.
2) Genetic algorithm for rapid optimization
When there are multiple strategy choices, how to choose the best goal or strategy? Mathematical methods based on linear programming or dynamic programming have slow calculation speed. In many cases, modeling is impossible or the amount of calculation is too large to calculate. This is genetic algorithm where it comes into play.
3) Neural network technology
The use of neural networks for autonomous driving training is the latest research hotspot. It is often said that letting neural networks learn to drive like humans is an exciting goal.
However, the problem with the neural network is that it is opaque, a black box system, and inexplicable. Basically, you cannot tell why the value of a node in the training model is 0.1 instead of 0.5. This is determined by the characteristics of the neural network. In addition, it is questionable whether a good model trained with training data can play the same role in the new environment.
One of the simplest ways to design an unmanned driving system is to use only neural networks for all control. We only need to train it with a large amount of data, so that there is no need to write complex control strategy algorithm codes, we only need to train A good neural network, and then use very little code to make it run, but before the interpretability of the neural network cannot be solved, the automatic driving system based entirely on the neural network is obviously not convincing and comfortable
Therefore, the control strategy based on inference logic in the unmanned driving system is still very important. It is the most feasible direction to make the white box control system based on inference logic and the black box control system based on neural network work together.
In addition to traditional pid control, vehicle control system technology has increasingly introduced neural network fuzzy control in unmanned vehicle systems.
Common driving decision control strategies for unmanned driving (automatic driving) systems
This is a traditional unmanned driving control strategy. The global route navigation uses the conventional A* algorithm (others can also be used). The image of the vehicle camera is processed by the neural network to extract the lane traffic signs and vehicle pedestrian information, and then use this information as input, use the vfh/dwa algorithm for real-time driving and obstacle avoidance control, such as lane change, deceleration and braking, etc.
Global planning still uses the same algorithm as strategy 1, but in real-time driving and obstacle avoidance, neural network technology is completely used, and the original pixel map features captured from the camera are used as the input of the neural network, and the output is for the car. We no longer extract lane information, traffic signs, vehicle and pedestrian signs from the original pixel map, all of which are handed over to the neural network for automatic identification, and the final output is steering and tracking, deceleration and braking, and other vehicle control commands.