Automated Vehicle Deadlock Prediction with Context-Aware Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automated vehicles face challenges in interpreting and resolving situations with unclear traffic conditions or low confidence levels in object detection, leading to deadlock situations that require external intervention, which existing technologies struggle to predict and manage effectively.
Innovation Solution
A method for predicting deadlock situations in automated vehicles by analyzing historical data to identify causes and contextual factors, such as seasonal patterns and event schedules, to generate predictive maps that indicate high-probability deadlock areas, enabling better resource planning and operational mode changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If automated vehicles use onboard sensor systems to model and interpret their surroundings, then the vehicles can achieve autonomous or semi-autonomous driving, but they may encounter deadlock situations when unable to interpret unclear traffic conditions or resolve situations with low confidence levels
Solution Approach 1:
The patent introduces a control center as an intermediary between automated vehicles and remote operators. The control center receives context information from vehicles, analyzes it using historical data and machine learning models, and determines whether to activate tele-operated driving modes. This intermediary layer resolves deadlocks by providing an additional decision-making level that can interpret unclear situations better than individual vehicle sensors alone.
Solution Approach 2:
The system performs preliminary actions by collecting and analyzing historical deadlock situation data before actual deadlock occurs. Machine learning models are trained in advance on historical context information to predict potential deadlock situations. This preliminary analysis enables the system to proactively identify and prepare for deadlock scenarios, reducing their occurrence through preventive measures.
2Reliability
If tele-operated driving is used to resolve deadlock situations, then external intervention can solve unclear traffic conditions, but communication delays and bandwidth limitations may affect response time and efficiency
Solution Approach 1:
The control center performs preliminary analysis of context information using pre-trained machine learning models before actual deadlock occurs. Historical data is analyzed in advance to establish patterns and predictors of deadlock situations. This preliminary preparation enables faster real-time decision-making when actual situations arise, reducing communication delays by having analysis frameworks ready beforehand.
Solution Approach 2:
The system applies different levels of remote intervention based on local situation requirements. Not all situations require full tele-operated driving activation. The control center can provide localized assistance through targeted instructions or information while allowing vehicles to maintain autonomous operation in less critical situations, thereby reducing unnecessary communication overhead and time loss.
3Productivity
If historical deadlock data is collected and analyzed to predict future deadlock situations, then resource planning can be improved, but data processing complexity and computational requirements increase
Solution Approach 1:
The data processing system is segmented into modular components: data collection modules at vehicle level, preliminary filtering and feature extraction at control center level, and advanced machine learning analysis for pattern recognition. This segmentation allows distributed processing that reduces overall system complexity while maintaining comprehensive analysis capabilities for improving resource planning.
Solution Approach 2:
The system uses machine learning models that create simplified representations (copies) of complex historical deadlock patterns. Instead of processing all raw historical data in real-time, the trained models store condensed knowledge representations that can quickly predict future deadlocks. This copying approach reduces computational complexity while preserving essential predictive capabilities.
Data Source
Figure 1
Figure 2
AI summary
Embodiments provide a method, a computer program, an apparatus, a vehicle, and a network entity for predicting a deadlock situation for an automated vehicle. The deadlock situation is an exceptional traffic situation necessitating a change of an operational mode of the vehicle (100). The method (10) for predicting a deadlock situation for the automated vehicle (100) comprises obtaining (12) information related to a historical deadlock situation cause and determining (14) historical context information for the historical deadlock situation cause. The method (10) further comprises determining (16) information indicative for the deadlock situation cause based on the information related to the historical context information and monitoring (18) actual context information for information indicative for the deadlock situation. The method (10) further comprises predicting (20) the deadlock situation based on the information indicative for the deadlock situation in the actual context information.