ResoSense: Automotive Bus Data Visualization
ROS-based platform for playback, decoding, and 3D visualization of CAN/CANFD bus data, LIDAR, and video streams for a 177k-employee automotive supplier.
The Challenge
A leading Tier 1 automotive supplier (177k employees, Canada) needed a unified platform for playback, decoding, and visualization of automotive bus and sensor data. Engineers were working with multiple file formats (MDF4, ADTF dat, rosbag), various communication protocols (CAN, CANFD, SomeIP, TAPI), and high-resolution sensor streams including LIDAR and video — all requiring efficient real-time processing for debugging and analysis.
Our Approach
We developed ResoSense, a ROS-based automotive data analysis platform that consolidates playback, decoding, and visualization into a single integrated tool.
Multi-Format Playback
A robust C++ playback engine handles MDF4/MF4, ADTF dat, and rosbag files. Optimized for large datasets, it enables engineers to replay test data and simulations with precise timing and synchronization.
Protocol Decoding
A custom C++ decoding library supports CAN, CANFD, SomeIP, and TAPI protocols. Using DBC, FIBEX, and ARXML database files, raw bus data is transformed into human-readable signals integrated into ROS data flows for real-time analysis.
GPU-Accelerated Visualization
Video and LIDAR sensor streams are processed using CUDA for hardware-accelerated decoding. RViz provides 3D visualization of point clouds, camera feeds, and decoded bus signals in a unified view.
Technical Stack
- Framework: ROS
- Languages: C++, Qt
- Protocols: CAN, CANFD, SomeIP, TAPI
- GPU: CUDA for video/LIDAR processing
- Visualization: RViz
- Testing: GTest
Results
- Unified playback, decoding, and visualization for automotive data
- Significant improvement in debugging and analysis workflows
- Support for all major automotive bus protocols
- Real-time LIDAR and video visualization via RViz