Adaptive Traffic Situation Prediction Under Sensor Failure
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current driver assistance systems and autonomous driving technologies face challenges in accurately assessing and predicting traffic situations, especially under conditions where certain sensors fail, such as poor lighting, requiring manual configuration or lack of redundancy.
Innovation Solution
An adaptive system using a multi-layer artificial neural network is trained on a wide range of sensor data from both inside and outside the vehicle, allowing it to select and combine available sensor data for accurate traffic situation determination and prediction, even without primary sensor data, through supervised or reinforcement learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If automated prediction of traffic situations is based on sensor data available at the current time, then the situation interpretation can be performed autonomously in the vehicle, but the system fails when primary sensors become unavailable due to poor lighting conditions or sensor malfunctions
Solution Approach 1:
The system performs preliminary actions by capturing and storing sensor data before the primary sensor fails. The buffer memory stores recent sensor data from multiple sensors, which can be retrieved and used for situation interpretation when the primary sensor becomes unavailable, ensuring continuous autonomous operation.
Solution Approach 2:
The system introduces an intermediary mechanism - a buffer memory that stores sensor data from multiple sources. When the primary sensor fails, this intermediary buffer provides alternative sensor data to the situation interpretation unit, maintaining system reliability without requiring manual intervention.
2Reliability
If compensation using other sensors is manually configured, then sensor failure can be compensated, but considerable manual effort and time are required for configuration
Solution Approach 1:
The system implements self-service by automatically selecting and switching to alternative sensors when the primary sensor fails. The situation interpretation unit continuously monitors sensor availability and automatically compensates for sensor failure without requiring manual configuration, eliminating time loss while maintaining reliability.
Solution Approach 2:
The system applies dynamics by making the sensor configuration adaptive and changeable in real-time. The situation interpretation unit dynamically adjusts which sensors are used based on their current availability and performance, allowing the system to automatically respond to sensor failures without fixed manual configuration.
3Ease of manufacture
If an engineer decides during development which systems use which sensors, then the sensor arrangement can be optimized for specific functions, but the system lacks adaptability when sensors fail or environmental conditions change
Solution Approach 1:
The system combines static optimization with dynamic adaptability. During development, the sensor arrangement is optimized for specific functions, but during operation, the situation interpretation unit dynamically adjusts sensor usage based on real-time conditions and sensor availability, providing both manufacturing ease and operational adaptability.
Solution Approach 2:
The system applies universality by designing the situation interpretation unit to handle multiple sensor types and configurations. The unit can interpret situations using various combinations of sensors, making the system universally adaptable to different sensor failures and environmental conditions while maintaining the optimized sensor arrangement.
Data Source
AI summary
The aim is to allow a reliable interpretation of a current or future traffic situation. To this end, a method for ascertaining a prescribed traffic situation is provided. The prescribed traffic situation is ascertained by a system (1) capable of learning that is trained to use first sensor data of a multiplicity of sensor data (9 to 16) from sensors inside and outside a motor vehicle (8) to detect the prescribed traffic situation. During operation, the first sensor data are then not provided, since the applicable sensors become inoperative, for example. The system capable of learning then automatically selects a portion of those sensor data from the multiplicity of sensor data (9 to 16) that are actually provided. Finally, the prescribed traffic situation is ascertained on the basis of the selected portion of the sensor data.