Jetson Nano ECU — Vehicle Control Stack (Team02-Course submodule)

This repository contains the core software that runs on an NVIDIA Jetson Nano used as an ECU in an autonomous driving lab platform. It is a submodule of the Team02-Course monorepo and provides perception, control, middleware, and tooling required to operate the vehicle in manual and autonomous modes.

Key features:

• Combined Controller: manual control via Xbox controller, PID/MPC steering, and speed PID control.
• Vehicle System: vehicle state model and actuation interface.
• Middleware: CAN <-> Zenoh bridge for battery, lights, speed, etc.
• Perception: lane detection, object detection, ACC, LKAS, traffic light/sign classification.
• Tools: camera calibration, Zenoh router (cloud bridge with InfluxDB), PID calibration, system monitor, VSS helpers.

Architecture:

• IPC is Zenoh-based (pub/sub). Middleware bridges vehicle CAN to Zenoh topics.
• Controllers switch between manual (SAE_0/SAE_1) and autonomous levels, consuming perception outputs to steer and modulate speed.
• Object detection triggers emergency braking and supports ACC (distance/speed tracking).
• Lane detection provides midpoint error for PID and trajectory for MPC.

Executables

• Vehicle System (src/vehicle.cpp)
  • Holds vehicle state, VSS-aligned signals, and actuates chassis/body.
• Middleware (src/middleware.cpp)
  • Publishes CAN-derived signals (battery, lights, etc.) and converts received CAN to Zenoh (e.g., speed).
• Combined Controller (src/combinedControl.cpp)
  • Threads:
    • ManualControl: reads Xbox inputs, exposes manual speed/steering.
    • PIDController or MPCController: steering control (mode-dependent).
    • SpeedPidController: regulates throttle using desiredSpeed/currentSpeed.
  • Uses manual inputs in manual modes, and perception-derived signals in autonomous modes.
• Detection (src/detection.cpp)
  • Lane Detection: midpoint error for PID and trajectory for MPC.
  • Object Detection: obstacle/road checks, emergency brake, ACC (relative speed), LKAS helpers.
  • Traffic lights/signs: crop and classify; results shared with controller to adjust speed/brake.

Communication

• Zenoh: primary IPC between components and tools.
• CAN Bus: vehicle-side communication; Middleware bridges to Zenoh.
• VSS: used to structure signals and naming.

Repository layout

• include/, src/: core code (ADAS, Controllers, Communication, Detection, Vehicle, …)
• config/
  • Zenoh/*.json: component configs
  • Systemd/*.service: systemd units for deployment
• tools/: calibration, PID tuning, system monitor, VSS, zenoh-router
• docs/: diagrams and documentation assets
• deploy/: Dockerfiles and scripts for Jetson deployment

Build

Prerequisites:

• Jetson Nano (JetPack 4.6+), CUDA (if using perception), OpenCV, Zenoh C/C++ libs, CMake 3.16+

Build:

• On device (recommended):
  • mkdir -p build && cd build
  • cmake ..
  • make -j
• For deployment via systemd, see config/Systemd and deploy/scripts.

Run (examples)

• Vehicle System:
  • ./build/VehicleSystem
• Middleware:
  • ./build/MiddleWare
• Combined Controller:
  • ./build/CombinedController
• Detection:
  • ./build/Detection

Systemd units (optional):

• config/Systemd/*.service contain ready-to-use services for each component.

Configuration

• Zenoh configs: config/Zenoh/*.json
• Environment and runtime settings: .env/, config/* (per-component)
• Camera calibration and model parameters: tools/cam_calibration/

Tools

• zenoh-router: bridge to cloud (InfluxDB plugin) for telemetry
• cam_calibration: camera intrinsic/extrinsic calibration helpers
• pid_calibrator: PID gain tuning
• system_monitor: runtime resource tracking
• vss, zenoh-router helpers

License

MIT — see LICENSE.

Team

Developed by Team02 at SEAME Portugal.