Consistency Evaluation System for ADAS ECU and Meter Display

Hiroyuki Takano (Nissan Motor)･Anhtuan Ngo･Ducphong Nguyen (Nissan Automotive Technology Viet Nam)･Tomohiro Kida (Fudo Giken Industry)･Daisuke Maruta (Accenture Japan)･Shigenori Tsunekado (Nissan Motor)

We developed an HIL system that automatically evaluates the consistency between ADAS ECU signals and meter displays. Icon images captured by a camera are recognized through image recognition and compared/judged in synchronization with the ECU signals. By converting the diverse and complex meter displays into signals and synchronizing them with the ECU signals for comparison, the evaluation process was simplified. This system was introduced into the vehicle verification process, where it was evaluated and contributed to reducing test man-hours.

A Method for Time Error Correction in SIL Simulation

Kyungnam Kim･Seungkyu Yoo･Hyunrak Ko･Sumin Park (Hyundai Autoever)

Precise time synchronization is crucial in Software-in-the-Loop (SIL) simulations. Conventional Sleep () functions lack microsecond accuracy, while Spinlock causes severe CPU degradation. This study proposes a hybrid time-delay method achieving 1-microsecond precision while maximizing CPU efficiency. It dynamically combines a low-cost timer (Sleep) with a high-cost timer (Spinlock). Using Welford's online algorithm, the system calculates the Sleep timer's statistical execution error in real-time. To save resources, it utilizes Sleep initially, transitioning to Spinlock only when the remaining wait time falls below the predicted error, effectively eliminating time discrepancies.

Research on Interface Methods between Virtual IP Models and Simulators in SIL Environments

Jaekyung Kim･Kyungnam Kim･Subin Jung･Haram Park (Hyundai Autoever)

In SIL environments, integrating virtual IP models with simulators using FMU standard causes performance degradation. This paper proposes direct interface approaches without FMU wrapper, defining optimized interfaces for data exchange and time synchronization. Experiments demonstrate reduced latency and increased throughput compared to conventional FMU-based approaches.

Development of a Parallel Level 4 Virtual Electronic Controller Unit Verification Platform Based on Cloud Infrastructure

Daeki Kim･YoungMo Lee (Hyundai Autoever)･HanSeung Myoung (Hyundai Motor)･Sunghoon Woo (Hyundai Autoever)

Conventional on-premises Level 4 virtual ECU verification has bottlenecks due to limited computing resources. This research proposes a parallel verification platform using cloud infrastructure. The platform executes multiple virtual ECU instances simultaneously in a distributed computing environment and processes verification in parallel according to software requirements. This reduces verification execution time and enables collaborative verification across multiple countries and departments. The results are expected to improve verification efficiency in automotive software development and shorten the development, verification, and deployment cycle.

Intelligent Fault Injection Scenario Generation Using Level 4 Virtual Electronic Controller Unit and Verification Efficiency Enhancement

YoungMo Lee･Daeki Kim･SungHoon Woo (Hyundai Autoever)

Manual fault injection scenario in Level 4 virtual ECU verification causes omission, redundancy, and inefficient resource allocation. This research proposes a methodology to automatically generate fault injection scenarios from functional safety requirements using AI. The methodology analyzes ISO 26262 requirements through machine learning and generates scenarios to enhance verification coverage. A priority determination algorithm eliminates redundancy and optimizes verification sequence, reducing verification time while expanding coverage. The results are expected to automate functional safety-oriented Basic Software verification, improve development processes, and reduce human errors.

Development of a TC39x TriCore Instruction Set Simulator for AUTOSAR OS QTDD ELF Verification Using AI-Assisted Methodology

Sunghoon Woo･Daeki Kim･Youngmo Lee･Kyungnam Kim･Jaekyung Kim (Hyundai Autoever)

This paper investigates whether AUTOSAR OS test ELFs can be executed on a SystemC ISS for the Infineon TC39x using AI-Assisted Methodology. The ISS implements six TriCore v1.6.2P cores with CSA chain, trap/interrupt handling, and 30 peripheral stubs (STM, MPU, MULTICAN, CAN-XL, GETH, FlexRay, ASCLIN etc). Completeness was improved via AI-driven datasheet analysis, MCAL sample code-based hardware modeling, and AI-assisted AUTOSAR OS boot register implementation. Key challenges -CSA linked-list reconstruction and 2-/6-core boot synchronization- were resolved under human knowhow.