Autonomous lane changing method and system fused with deep reinforcement learning

A reinforcement learning and in-depth technology, applied in neural learning methods, biological neural network models, neural architectures, etc., can solve problems such as inability to achieve autonomous lane changes, endanger the safety of drivers and passengers, and affect urban traffic, so as to ensure the accuracy of decision-making. , improve applicability, improve the effect of decision-making speed
CN113682312APending Publication Date: 2021-11-23中汽创智科技有限公司
5 Cites 3 Cited by

Patent Information

Authority / Receiving Office
CN · China
Current Assignee / Owner
中汽创智科技有限公司
Publication Date
2021-11-23

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention discloses an autonomous lane changing method and system fused with deep reinforcement learning. The method comprises the steps: training a deep reinforcement learning model in a training environment, and obtaining and storing training parameters; in the training environment, information of a target vehicle driving according to a rule-based automatic driving strategy is added; formulating an evaluation function according to the training environment and the rule-based automatic driving strategy; judging whether the information of the target vehicle meets an arbitration condition or not according to the evaluation function; if yes, the training parameters are fused into information of the target vehicle, and the target vehicle is controlled to run; and if not, the target vehicle is still controlled to run according to the rule-based automatic driving strategy. According to the method, deep reinforcement learning and a rule-based automatic driving strategy are fused, a large amount of work of traversing a lane changing scene for modeling is omitted in an unmodeled environment, and applicability, decision accuracy, decision efficiency and driving safety are improved.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention relates to the technical field of automatic driving decision-making planning, in particular to an autonomous lane-changing method and system integrated with deep reinforcement learning. Background technique

[0002] In recent years, the rapid development of automatic driving technology has brought great convenience to people's life and work; and high-level automatic driving functions, such as autonomous overtaking, automatic assisted navigation and driving, etc., all require the sub-function of autonomous lane change. The complex and changeable urban traffic conditions have brought great challenges to the development of autonomous lane changing.

[0003] The current mainstream approach to this problem is to define different scenarios by formulating rules, formulate different lane-changing algorithms and parameters in different scenarios, and make vehicles follow the planning based on the information of detected adjacent vehicles and traff...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More