Vehicle Perception Model Selection for Road-Specific Autonomous Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vehicle control methods face challenges in achieving accurate perception due to the limited capacity of single perception models, which can lead to perception failures, especially in diverse real-world situations and environmental conditions, affecting driving safety and public trust in autonomous driving.
Innovation Solution
A slice-based perception approach is implemented, where perception models are trained for specific road segments and environmental conditions, forming a perception model library, allowing for the dynamic selection and use of the most adapted perception model based on real-time vehicle data, enhancing perception accuracy and robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single perception model is used for all road segments and environmental conditions, then the device complexity is reduced, but the measurement precision and reliability of perception deteriorate
Solution Approach 1:
The patent divides the perception model into multiple specialized models, each trained for specific road segments and environmental conditions. Instead of using one general model, the system segments the perception task into multiple specialized models that can be selected based on current driving conditions, thereby improving perception accuracy without significantly increasing overall system complexity.
Solution Approach 2:
The patent applies local quality by creating perception models with specialized characteristics tailored to specific road segments and environmental conditions. Each model has optimized features and parameters suited for its designated scenario, allowing the system to achieve high perception accuracy for diverse conditions while maintaining manageable model complexity through targeted specialization.
2Ease of operation
If a single perception model is used for all situations, then the ease of operation is improved, but the reliability of vehicle control deteriorates
Solution Approach 1:
The patent implements a dynamic model selection mechanism that automatically chooses the appropriate perception model based on real-time detection of road segment characteristics and environmental conditions. This dynamic adaptation ensures reliable vehicle control by always using the most suitable model for current conditions, while the automated selection process maintains ease of operation without requiring manual intervention.
Solution Approach 2:
The system employs feedback mechanisms where perception results and environmental sensor data are continuously monitored to determine the current road segment and condition type. This feedback loop enables automatic selection of the appropriate specialized perception model, ensuring reliable control while maintaining operational simplicity through automated decision-making based on real-time conditions.
Data Source
AI summary
A vehicle control method performed by a computer device includes obtaining traveling information of a target vehicle that indicates a target road segment that the target vehicle to travel on and a target environmental condition when the target vehicle is on the target road segment, obtaining a target perception model of the target road segment under the target environmental condition from a perception model library that stores perception models of a plurality of road segments under different environmental conditions, calling the target perception model to perform a perception task on the target road segment to obtain a perception result, and controlling, based on the perception result, the target vehicle to travel on the target road segment.