APP4MC RaceCar: A Practical ADAS Demonstrator for Evaluating and Verifying Timing Behavior
summary
A model-based RC demonstrator (APP4MC) runs perception and planning across Jetson AGX + BeagleBone AI with ZED2 camera, Hokuyo LiDAR, VESC motor control and CAN/UART links, using RT Linux and a gB RTOS. It generates real-time traces via the BTF tracing framework (visualized in Eclipse Trace Compass) and extracts end-to-end latencies into an MLTrace-style database; current status uses software tracing and NVIDIA CUDA profilers, tasks are statically pinned and failover is not implemented yet.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:40Agenda
- 5:10Existing ADAS Platforms
- 6:30APP4MC
- 8:53System Architecture
- 9:32Application Stack
- 10:57Architecture
- 12:27Timing Analysis
- 14:08Conclusion
- 15:13Questions
related talks
28mOpenAdx "Autonomous Driving Simulation" Testbed – hardware in the loop
35mGetting started with a self-driving RC car
31mAutonomous driving, how Python drives Formula Racing cars
23mRacing Robocars
06mEmbedded Executive LIVE
35mDingoCar: Perception module Small scale, self-driving car
20m96Boards - Autoware: 1 year on
19mBoosting confidence in tiny AI deployment in embedded control systems