Objective
To support Level 3+ autonomous driving use cases, a German Tier-1 supplier collaborated with AVIN Systems for developing and integrating an autonomous driving framework. The project included AI-based perception, fusion, and path planning algorithms integrated into the ECU platform.
Solution
Requirement Analysis & Configuration
Defined AI-based architecture for perception, fusion, and path planning and Integrated with Adaptive AUTOSAR.
Network Management and Development
Built ROS2-based communication modules for real-time decision-making.
Memory and Bootloader Integration
Enabled secure OTA updates and persistent data logging on NVIDIA Xavier.
Security Enhancement
Implemented encrypted data channels and secure boot mechanisms.
Testing and Validation
Validated ML algorithms using simulation and HIL testing and Ensured ISO 26262 compliance.
Impacts
- Improved perception accuracy by 15% using ML-based fusion.
- Enabled real-time decision-making under latency constraints.
- Delivered modular, safety-certified software for scalable deployment.
Tech Stack
- Python
- C++
- TensorFlow
- ROS2
- AUTOSAR Adaptive
- NVIDIA Xavier
- QNX
- Linux.