Method for optimizing decision-making regulation and control, method for controlling vehicle traveling, and related devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous driving systems face challenges in further improving traveling safety, as existing decision-making and control methods do not adequately optimize the interaction between behavior decision-making and motion planning layers, leading to suboptimal vehicle trajectory and behavior outputs.
Innovation Solution
A closed-loop optimization method is introduced, where both the behavior decision-making and motion planning layers are optimized based on differences between actual and teaching sequences, using a determining model to adjust parameters and improve the system's performance, ensuring the vehicle's trajectory and behavior align with safe and optimal teaching sequences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional feedback optimization is used where motion planning layer provides feedback to behavior decision-making layer, then vehicle traveling safety is improved to some extent, but the optimization is incomplete and traveling safety needs further improvement
Solution Approach 1:
The optimization system is segmented into two independent optimization modules: a first optimization module for the behavior decision-making layer and a second optimization module for the motion planning layer. Each module independently optimizes its respective layer by calculating loss functions and updating parameters, rather than using a single unified feedback loop. This segmentation enables more comprehensive optimization coverage while maintaining modular system architecture.
Solution Approach 2:
The system implements dual feedback mechanisms: (1) traditional feedback from motion planning layer to behavior decision-making layer, and (2) new feedback from the independently optimized behavior decision-making layer back to motion planning layer. This bidirectional independent feedback ensures both layers are fully optimized and can further improve traveling safety by addressing limitations of unidirectional feedback.
2Manufacturing precision
If behavior decision-making layer and motion planning layer are optimized independently through dual optimization modules, then comprehensive parameter optimization is achieved, but computational complexity increases
Solution Approach 1:
The computational optimization process is divided into separate segments: the first optimization module computes loss functions and updates parameters for the behavior decision-making layer, while the second optimization module independently does the same for the motion planning layer. This segmentation allows each module to focus computational resources on its specific layer, achieving precise parameter optimization without requiring simultaneous complex computations across the entire system.
Solution Approach 2:
The system performs preliminary optimization actions by first independently optimizing the behavior decision-making layer through its dedicated optimization module, then using the optimized behavior outputs as inputs for the motion planning layer optimization. This sequential preliminary optimization reduces the overall computational complexity compared to simultaneous optimization of both layers, as each optimization can proceed with fixed inputs from the other.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A method for optimizing decision-making regulation and control, a method for controlling traveling of a vehicle by using a decision-making and control system, and a related apparatus in the field of artificial intelligence autonomous driving are provided. In the method for optimizing decision-making regulation and control, a first traveling sequence is obtained, where the first traveling sequence includes a first trajectory sequence of the vehicle in information about a first environment and first target driving behavior output by a behavior decision-making layer of a decision-making and control system based on the information about the first environment. A second traveling sequence is obtained, where the second traveling sequence includes a second trajectory sequence output by a motion planning layer of the decision-making and control system based on preset second target driving behavior and the second target driving behavior. The behavior decision-making layer is optimized based on a difference between the first traveling sequence and a preset traveling sequence, and the motion planning layer is optimized based on a difference between the second traveling sequence and the preset traveling sequence. Closed-loop optimization of the behavior decision-making layer and the motion planning layer can be implemented by using the method for optimizing decision-making regulation and control, to ensure that the vehicle can travel normally.