ODD Validation for Autonomous Driving Safety
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous driving systems face significant challenges in verifying safety and reliability due to the combinatorial explosion of possible scenarios, leading to costly and time-consuming validation and high hardware requirements, particularly for perception systems and lane-keeping features.
Innovation Solution
A control system that determines the operational design domain (ODD) based on road safety barrier and road characteristics metrics, allowing the driver support function to be activated only when these conditions are met, thereby reducing the need for high-end sensors and relaxing integrity requirements for lane-keeping features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional testing methods are used to verify safety and reliability of autonomous driving systems, then the system can meet safety standards, but the validation process becomes extremely costly and time-consuming due to combinatorial explosion of scenarios
Solution Approach 1:
The patent segments the operational design domain into discrete road types and lane types with specific characteristics. By dividing the continuous space of possible driving scenarios into categorized segments (e.g., highway, urban road, rural road with specific lane configurations), the system can systematically validate each segment type rather than testing all possible scenarios, thereby reducing validation complexity while maintaining comprehensive safety coverage
Solution Approach 2:
The patent performs preliminary classification of the operational environment into defined road and lane types before validation. By pre-defining categories such as 'highway with divided carriageway' or 'urban road with one-way traffic', the system establishes a structured framework that guides subsequent validation efforts, avoiding the need to handle combinatorial explosion during actual testing
2Reliability
If high safety requirements are imposed on the perception system to ensure accurate lane position determination, then the system achieves high reliability, but significant redundancy and expensive sensors are required
Solution Approach 1:
The patent applies local quality by determining lane position accuracy requirements based on the specific road type and lane type. For example, on highways with divided carriageways and clear markings, the system can use standard sensors with relaxed accuracy requirements, while on complex urban roads with frequent lane changes, higher accuracy may be needed. This localized approach to quality requirements eliminates the need for uniform high-end sensors throughout all operating conditions
Solution Approach 2:
The patent changes the parameters of the perception system based on the operational context. By adjusting lane position determination thresholds, accuracy requirements, and sensor activation levels according to the classified road and lane types, the system dynamically adapts its performance parameters to match the actual risk and complexity of the driving environment, reducing unnecessary hardware requirements
3Adaptability or versatility
If the driver support function is made available in all road conditions, then the system achieves high versatility, but the integrity requirements and validation complexity increase significantly
Solution Approach 1:
The patent makes the availability of the driver support function dynamic based on the classified operational design domain. The system continuously evaluates the current road type and lane type against predefined criteria, enabling or disabling the driver support function in real-time. This dynamic adaptation allows the system to achieve high versatility in supported road conditions while maintaining manageable integrity requirements by restricting operation to well-defined, validated scenarios
Data Source
Figure 1a~1b
Figure 2
Figure 3
AI summary
A control system for a vehicle traveling in a first direction on a road segment is presented. The vehicle has a driver support function for autonomously manoeuvring the vehicle, and the driver support function is capable of operating within an operational design domain (ODD) including a road safety barrier metric and a road characteristics metric. The control system comprises control circuitry configured to obtain data comprising information about a surrounding environment of the vehicle. The information includes road safety barrier data and road characteristics data. Further, the control circuitry is configured to determine a fulfilment of the ODD based on the obtained data by determining a fulfilment of the road safety barrier metric and the road characteristics metric based on the road safety barrier data and the road characteristics data, respectively. Moreover, if the ODD is fulfilled, the control circuitry is configured to generate a first control signal in order to control an availability of the driver support function for the road segment so to make the driver support function available for an occupant of the vehicle.