Autonomous Vehicle U-Turn Strategy Determination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies lack a systematic approach to determine an optimal U-turn strategy for autonomous driving vehicles, which is crucial for safe navigation and accident prevention during U-turns, as they do not effectively consider various situational factors.
Innovation Solution
A device and method that utilize deep learning to categorize situation information relevant to U-turns, such as collision avoidance with preceding or neighboring vehicles, pedestrians, and traffic lights, to determine the best U-turn strategy based on learned models and adjust scores based on risk factors like collision warnings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If deep learning is applied to determine U-turn strategy by categorizing situation information, then the accuracy and safety of U-turn navigation is improved, but the system complexity and computational requirements increase
Solution Approach 1:
The patent segments situation information into distinct categories (preceding vehicle information, neighboring vehicle information, pedestrian information, traffic light information, and road information) and processes each category separately through dedicated processing modules. This segmentation allows the complex deep learning system to handle different types of information in an organized manner, improving reliability without overwhelming system complexity.
Solution Approach 2:
The patent introduces a scoring dimension to evaluate multiple U-turn strategies (first U-turn strategy, second U-turn strategy, third U-turn strategy) by assigning scores based on learned situation information. This adds an evaluation dimension that systematically compares different strategies, enhancing safety through quantitative assessment while maintaining manageable system complexity through structured scoring.
2Productivity
If multiple situational factors are considered and dynamically adjusted, then the optimality of U-turn route is improved, but the computational time and processing load increase
Solution Approach 1:
The patent performs preliminary categorization and organization of situation information into predefined groups before deep learning processing. By pre-structuring the input data according to established categories (vehicle positions, pedestrian locations, traffic light states), the system reduces computational complexity during the actual U-turn decision-making process, thereby improving route optimality without excessive computational time.
Solution Approach 2:
The patent implements a feedback mechanism where the learned situation information and scoring results are used to dynamically adjust and select the optimal U-turn strategy. The system continuously evaluates multiple strategies, compares their scores, and selects the best one, creating a closed-loop feedback system that improves route optimality through iterative assessment rather than exhaustive computation.
Data Source
AI summary
A device and a method for determining a U-turn strategy of an autonomous driving vehicle is disclosed. The device for determining a U-turn strategy of an autonomous driving vehicle includes a learning device that learns a U-turn strategy for each situation by dividing situation information to be considered in a U-turn of the autonomous driving vehicle into groups, and a controller that determines the U-turn strategy of the autonomous driving vehicle based on the U-turn strategy for each situation learned by the learning device.