USE CASE: Cooperative perception for driving assistance
In the rapidly evolving field of automated driving, the SmartEdge project is at the forefront of developing cutting-edge technologies for test case generation. By integrating predefined knowledge with real-time traffic situations, SmartEdge aims to enhance the quality, completeness, and consistency of test cases. This blog post highlights the challenges of creating test cases for Advanced Driver-Assistance Systems (ADAS) and introduces the innovative approach proposed by SmartEdge.
Challenges in ADAS Test Case Generation
Generating accurate and comprehensive test cases for ADAS involves capturing the behaviour of various road users using diverse and heterogeneous data sources. This includes information from ego-vehicles, road infrastructure, satellites, images, and vehicle dynamics. Processing such complex and varied data poses significant challenges in extracting useful knowledge for ADAS testing. Additionally, the vast number of real-world driving scenarios makes it difficult to ensure complete coverage and accurate representation of all corner cases.
Leveraging Expert Knowledge and Data
To address these challenges, SmartEdge explores the combination of large, heterogeneous data streams with expert domain knowledge provided by test engineers and ADAS developers. The project proposes representing expert knowledge and abstract situations using semantic-based declarative languages. By employing reasoning systems, SmartEdge aims to construct exhaustive sets of test cases for ADAS. This approach allows for a comprehensive representation of various scenarios that ADAS should handle.
Real-Time Data Integration for Realistic Test Cases
SmartEdge combines abstract knowledge with real-time data to create realistic test cases for on-the-road scenarios. By leveraging the vehicle’s received data and predefined traffic rules stored as abstract domain knowledge, the system estimates the compatibility of the current data with traffic rules. This enables the generation of test cases that align with real-world conditions, without the need for complex testing environments. The technology builds upon SmartEdge’s existing work, which provides a unified framework for semantic stream fusion [1,2].
In conclusion, the SmartEdge project represents a significant advancement in test case generation for ADAS. By combining expert knowledge with real-time data, SmartEdge aims to create comprehensive and realistic test cases to ensure the safety, robustness, and regulatory compliance of highly automated driving systems. Through the use of semantic-based languages and reasoning systems, SmartEdge is paving the way for more effective and efficient testing methodologies in the field of driving assistance.
Author: Trung Kien Tran, Lead Scientist at Bosch Center for AI, Germany.
 CQELS 2.0: Towards A Unified Framework for Semantic Stream Fusion. Anh Le-Tuan, Manh Nguyen-Duc, Chien-Quang Le, Trung-Kien Tran, Manfred Hauswirth, Thomas Eiter, Danh Le-Phuoc
 VisionKG: Towards A Unified Vision Knowledge Graph. Anh Le-Tuan, Trung-Kien Tran, Manh Nguyen-Duc, Jicheng Yuan, Manfred Hauswirth, and Danh Le-Phuoc