Scenario-Aware Perception System for Automated Vehicles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automated vehicles face difficulties in reliably identifying and tracking other vehicles due to unknown orientations, which complicates signal processing from sensors like cameras, radar, and lidar, especially in varying traffic scenarios.
Innovation Solution
A scenario-aware perception system that uses sensors and digital maps to determine the present traffic scenario and select a preferred algorithm optimized for that scenario, allowing for improved object detection and tracking by considering road geometry and anticipated vehicle maneuvers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a single fixed algorithm is used for object detection, then the system is simple to implement, but it cannot adapt to varying traffic scenarios and unknown vehicle orientations
Solution Approach 1:
The system dynamically selects from multiple detection algorithms based on the current traffic scenario. A scenario classifier analyzes sensor data to determine the present situation (e.g., intersection, highway, urban area) and selects the most appropriate algorithm from a library of scenario-optimized algorithms, allowing the system to adapt to varying conditions without being overly complex
Solution Approach 2:
The system changes operational parameters by selecting different algorithms optimized for specific scenarios. Each algorithm has parameters tuned for particular conditions (e.g., speed thresholds, detection ranges, orientation expectations), allowing the system to optimize performance for each scenario while maintaining a manageable overall structure through parameterization
2Reliability
If multiple optimized algorithms are maintained for different scenarios, then detection reliability improves, but system complexity and computational overhead increase
Solution Approach 1:
The system segments the detection problem into distinct traffic scenarios (intersection, highway, urban, rural, etc.), with each segment having its own optimized algorithm. This segmentation allows reliable detection within each scenario while managing overall complexity through modular organization of algorithms by scenario type
Solution Approach 2:
A scenario classifier acts as an intermediary between raw sensor data and the detection algorithms. This mediator analyzes current conditions, determines the active scenario, and selects the appropriate algorithm, thereby managing the complexity of having multiple algorithms while ensuring reliable detection through scenario-matched selection
3Productivity
If scenario-based algorithm selection is implemented, then processing speed and accuracy improve, but computational overhead for scenario classification increases
Solution Approach 1:
Algorithms are pre-configured and optimized for specific scenarios during system initialization or offline preparation. Scenario-specific parameters and thresholds are predetermined, so during operation the system only needs to classify the current scenario and apply the pre-prepared algorithm, reducing real-time computational energy consumption while maintaining high processing speed
Data Source
AI summary
A scenario aware perception system (10) suitable for use on an automated vehicle includes a traffic-scenario detector (14), an object-detection device (24), and a controller (32). The traffic-scenario detector (14) is used to detect a present-scenario (16) experienced by a host-vehicle (12). The object-detection device (24) is used to detect an object (26) proximate to the host-vehicle (12). The controller (32) is in communication with the traffic-scenario detector (14) and the object-detection device (24). The controller (32) configured to determine a preferred-algorithm (36) used to identify the object (26). The preferred-algorithm (36) is determined based on the present-scenario (16).