Plausibility Rules for Driving Assistance Confidence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current driving assistance systems face reliability issues due to detection inaccuracies and errors in sensor data, leading to potentially confusing or unacceptable system responses, especially when predicting the behavior of multiple entities or objects in complex scenarios.
Innovation Solution
A method for improving confidence estimates by applying plausibility rules to sensor data, which assess the logical consistency of detected entities and their relations, allowing for more reliable predictions and decision-making in driving assistance systems, even with less sophisticated or lower-quality sensor equipment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If additional sensor equipment and high-performance equipment are provided to enhance system reliability, then detection accuracy and prediction reliability are improved, but hardware complexity and costs increase
Solution Approach 1:
The patent changes the parameter of confidence estimation by introducing plausibility rules that evaluate the consistency of sensor data with expected physical and logical relationships. Instead of relying solely on sensor quality, the system evaluates whether detected entities and their relations make sense (e.g., vehicles maintaining plausible distances, consistent motion patterns), thereby improving prediction reliability through software-based parameter validation rather than hardware upgrades
Solution Approach 2:
The patent substitutes mechanical/sensor-based reliability improvement with an information-processing approach. By replacing the need for additional high-performance sensors with plausibility rule evaluation, the system uses logical reasoning and data consistency checks to achieve reliable predictions, substituting hardware complexity with software-based validation mechanisms
2Adaptability or versatility
If sensor data from multiple entities are used for sophisticated functions, then system functionality is improved, but accumulated inaccuracies lead to wrong decisions and unreliable predictions
Solution Approach 1:
The patent implements feedback through plausibility rules that continuously evaluate whether the combined sensor data from multiple entities maintains logical consistency. The system checks if detected entities and their relationships satisfy expected constraints (e.g., gap widths between vehicles, relative positions, motion coherence), providing feedback that filters out accumulated inaccuracies and prevents wrong decisions in sophisticated functions like lane change assistance
Solution Approach 2:
The patent introduces plausibility rules as an intermediary layer between raw sensor data and prediction decisions. This intermediary evaluates the consistency of multi-entity data before it influences system decisions, acting as a filter that prevents accumulated inaccuracies from propagating through the decision-making process, thereby maintaining reliability while enabling sophisticated functionality
3Ease of operation
If sensor uncertainty is not explicitly considered, then system operation is simpler, but detection inaccuracies and errors lead to confusing or unacceptable system responses
Solution Approach 1:
The patent applies preliminary action by evaluating plausibility rules before sensor data is used for predictions or control decisions. The system pre-validates the consistency of detected entities and their relationships against expected physical and logical constraints, filtering out unreliable data before it reaches the decision-making layer, thereby maintaining simple operation while ensuring reliable responses
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a driving assistance system (100) including a prediction subsystem (110) in a vehicle. According to a method aspect of the invention, the method comprises the steps of accepting an environment representation (120); calculating a confidence estimate related to the environment representation based on applying plausibility rules (122) on the environment representation, and providing the confidence estimate as input for an evaluation of a prediction based on the environment representation.