Adaptive control method for driving

Abstract

The invention discloses an adaptive control method for driving and belongs to the technical field of intelligent driving and artificial intelligence. The adaptive control method includes the steps that a historical driving data set is obtained and divided into a training set, a testing set and a verifying set; a deep reinforcement learning algorithm based on a deep convolution neural network is adopted for constructing a network model used for driving control; training set data are used for training the network model, gradient iteration of a cost function is used for repeatedly training the network model, and the optimized network model is obtained; the testing set and the verifying set are used for verifying the performance of the optimized network model, and the network model meeting theperformance is adopted as an adaptive decision model; and the adaptive decision model is used for processing real-time environment data collected currently, and a driving decision is made. By means of the adaptive control method for driving, the driving decision made by people in a true complex environment and corresponding driving actions can be better simulated.

Description

technical field [0001] The invention relates to the technical fields of intelligent driving and artificial intelligence, in particular to a driving adaptive control method. Background technique [0002] In recent years, with the rapid rise of intelligent driving technology and artificial intelligence technology, more and more universities, enterprises, and research institutes have begun experimental testing of intelligent vehicles, and have gradually shifted from basic technology research and development to market applications. The development of intelligent driving has fundamentally changed the traditional way of driving vehicles, liberating the driver from the complex driving environment, using technologies such as environmental perception, radar equipment, autonomous positioning, decision-making planning, and intelligent control to realize the Without human active interference, it can automatically, safely and efficiently complete human-like driving behaviors such as auto...

AI Technical Summary

Problems solved by technology

However, due to the complexity and unpredictability of actual application scenarios, the strategy based on expert rules may not cover a wide enough range of scenarios, which may easily cause serious traffic accidents

In addition, when the newly added rules conflict with the original rules, the original rules must be adjusted and modified, which greatly reduces the robustness and adaptability of the strategy based on expert rules.

Based on the control theory of deep learning, Nvidia built the mapping relationship between environmental information and vehicle control parameters by training a deep convolutional neural network in 2016 to achieve "end-to-end" control. Therefore, it relies too much on deep learning and lacks rational and objective analysis. At the same time, this method requires a large amount of labeled data sets, which greatly increases the workload of researchers.

Images