Vehicle Perception Control for Deviating Driver Behavior
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for assessing the reliability and safety of autonomous systems, particularly in mixed traffic situations with human-driven vehicles, are inadequate due to the complexity of interactions and the combinatorial explosion of possible scenarios, leading to costly and time-consuming validation processes.
Innovation Solution
A method and system that detect deviating vehicle behavior and properties using perception systems, classify them, and transmit information to other vehicles to enable them to prepare for potential interactions, allowing for proactive control measures such as adjusting speed or lane changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional testing methods are used to validate autonomous systems, then safety and reliability requirements can be enforced, but the validation process becomes costly and time-consuming due to the combinatorial explosion of possible situations
Solution Approach 1:
The system performs preliminary classification of vehicles into normal and deviating categories before actual interactions occur. By pre-identifying and categorizing deviating vehicles using perception systems and machine learning, the system prepares advance information about potential safety issues, allowing other vehicles to take preventive actions rather than reacting to each situation in real-time, thus reducing validation complexity and time
Solution Approach 2:
The patent introduces an intermediary classification system that acts as a mediator between raw sensor data and safety validation processes. This intermediary layer categorizes vehicles into standardized types (normal, deviating with specific subcategories), simplifying the complex validation process by providing structured intermediate representations that are easier to analyze and validate against safety requirements
2Reliability
If autonomous vehicles strictly follow traffic rules, then safety standards are met, but they cannot adequately handle informal human driving behavior that deviates from rules
Solution Approach 1:
The system dynamically adapts its behavior based on the classification of other vehicles. When a deviating vehicle is detected and classified, the autonomous vehicle modifies its standard rule-following behavior to account for the predicted actions of the deviating vehicle. This dynamic adjustment allows the system to maintain safety while adapting to informal human driving behaviors that deviate from standard traffic rules
Solution Approach 2:
The system changes operational parameters such as speed, distance to following vehicle, and lane change timing based on the classification of surrounding vehicles. When deviating vehicles are detected, the autonomous vehicle adjusts these parameters proactively (e.g., increasing following distance, reducing speed) to compensate for unpredictable human behavior, thereby maintaining safety while adapting to non-standard driving patterns
3Adaptability or versatility
If autonomous vehicles interact with human-driven vehicles in mixed traffic, then real-world applicability increases, but the complexity of predicting and responding to unpredictable human actions increases
Solution Approach 1:
The system segments the complex problem of handling mixed traffic by classifying vehicles into distinct categories (normal vehicles, deviating vehicles with specific subcategories). This segmentation breaks down the complexity of predicting all possible human behaviors into manageable categories, where each category has characteristic patterns that can be predicted and responded to with specific strategies, reducing overall system complexity while maintaining interaction capability
Data Source
AI summary
The present disclosure relates to a method to control a vehicle system. The method includes detecting a deviating vehicle having at least one of a deviating behaviour and a deviating vehicle property by means of a perception system of a first vehicle, wherein the perception system includes at least one sensor device configured to monitor a surrounding environment of the first vehicle; assigning a deviating vehicle classification to the deviating vehicle based on the deviating behaviour and/or deviating vehicle property; determining at least one second vehicle to receive information relating to the deviating vehicle; and transmitting to each determined second vehicle a set of information relating to the deviating vehicle, said set of information including at least one of the deviating vehicle classification and a predetermined instruction to be performed by the determined second vehicle, wherein the predetermined instruction is dependent on the deviating vehicle classification.