BFMC (RAVEN Version) - Documentation Project

This repo contains the sources for the Bosch Future Mobility Challenge documentation. Find it at this link

Please find here the Team Raven: Technical Dependency Map

---

🚦 Project Workflow & Dependencies

To manage our team effectively, we utilize a structured numbering system and parallel development tracks.

1. The Gap (Issues 009 – 018): The "Hardware Arrival" Phase

The gap in issue numbering is reserved for the Physical Car Integration phase once the DHL package arrives. While we are currently in a "No-Car" phase, these missing numbers represent future milestones for:

• Physical Assembly: Mounting the camera and sensors.
• Sensor Calibration: Moving from virtual camera math to real lens calibration.
• Real-World PID Tuning: Adjusting the and values on actual asphalt/tiles instead of Gazebo friction.
• Power Management: Testing the actual battery sag and voltage rails.

Task 019a serves as the bridge; it is the documentation audit performed now to ensure we are ready to fill the 009–018 gap the moment the car is unboxed.

2. Sub-task Dependencies (e.g., 008a vs. 008b)

In most cases, "a" must start before "b," but they can be worked on simultaneously once the "handshake" is defined.

• Sequential Dependency: For 008a (Capture) and 008b (AI), you cannot train the AI without data. One member takes screenshots (008a) to provide the dataset for the recognition model (008b).
• Parallel Development: The member on 008b can write the "Logic Skeleton" (code expecting an image and returning a label) using placeholder images while waiting for data from 008a.

3. Inter-Team Parallelism (e.g., 007X vs. 008X)

Work happens in parallel across our 3 Teams:

• Team C can finish the Dashboard (007a) using "Mock Data" (fake speed/steer numbers) to test the UI layout.
• Team A can be mid-development on the Sign AI (008b).
• The Connection: Teams only need to meet at the end to "plug" the real AI output into the Dashboard.

---

👥 Team Roster & Task Allocation

For a detailed breakdown of team roles and member contributions, please refer to our Official Membership & Tasks Sheet. All development tasks are distributed across the 5-member squad to ensure parallel progress.

Note on Version Control: To maintain a clean main branch, Hatem Soliman acts as the Integration Lead, reviewing and committing modules submitted by Team A, B, and C members. Thus, the commit history reflects the integration events rather than individual author timestamps.

📋 Kanban Board & Team Structure

Our GitHub Project board uses a specific labeling system to define ownership and task types.

Team Definitions

• Team A: Perception & Data (The "Eyes"): Focused on the vision pipeline, OpenCV, and AI models. They use raven-brain-stack (Vision) and raven-sim.
• Team B: Hardware & Low-Level (The "Spine"): Focused on the physical car, C++ Nucleo code, and PID control. They use raven-embedded-control and bfmc-documentation.
• Team C: Integration & Logic (The "Joker"): The decision-makers who connect Team A to Team B. They write the State Machine (FSM) and Dashboard in raven-brain-stack and raven-computer.

How to Read Issues

Each issue on the Kanban board follows this format: [Issue #] [Task Name] - [Team Label].

1. GENERAL Status: Foundational tasks (like 001a and 019a) that everyone must understand or perform to set up their environment.
2. SPECIALIZED Status: Technical features specific to a single team's mission (e.g., PID math for Team B or AI for Team A).

---

📝 Summary of Workflow Logic

• General Tasks (001a, 019a): Everyone starts here immediately to ensure the "Digital Twin" and "Hardware Map" are understood.
• Sub-tasks (a, b, c): These are "Micro-steps" within a feature. Usually, "a" provides the input for "b".
• Team Parallelism: Team A, B, and C work at the same time because the Serial Protocol acts as a contract between them.

🚀 Recently Documented Features

• Speed PID Controller (Task 003a):

  • Closed-loop velocity control for the Nucleo.
  • Tunable Kp, Ki, Kd parameters via serial.
  • Documentation: source/data/embeddedplatform/speedPid.rst

• Steering & MPC Control (Task 003b):

  • Combined Speed + Steering command (#mpc:speed;steer).
  • Anti-jitter logic for servo protection.
  • Documentation: source/data/embeddedplatform/steeringControl.rst

• IPM Matrix Calculation (Task 002a):

  • Tools for calculating Inverse Perspective Mapping (Birds-eye view).
  • Includes Mac/PC local development support.
  • Documentation: source/data/brain/ipm.rst

• Video Stream Handler (Task 001b):

  • Bridges Gazebo/ROS images to OpenCV.
  • Subscribes to /camera/rgb/image_raw and publishes preview.
  • Documentation: source/data/brain/videostream.rst

• Lane Segmentation (Task 002b):

  • Isolates the track using ROI masking and thresholding.
  • Publishes binary mask to /raven/perception/lane_mask.
  • Documentation: source/data/brain/lanesegmentation.rst

• Lateral Control (Task 002c):

  • Calculates steering error from lane mask centroid.
  • Publishes error to /raven/control/lateral_error.
  • Documentation: source/data/brain/lateralcontrol.rst

• Python Serial Class (Task 005a):

  • Robust 10Hz serial sender for Nucleo communication.
  • Thread-safe speed/steer updates and auto-reconnection.
  • Documentation: source/data/brain/seriallink.rst

• Steer Angle Mapping (Task 005b):

  • Converts lateral error to physical steering angles.
  • Documentation: source/data/brain/steering_mapper.rst

• FSM Framework & Sign Logic (Task 006a/b):

  • Central Brain managing states (Idle, Drive, Stop).
  • Enforces traffic rules (Stop Sign logic).
  • Documentation: source/data/brain/fsm.rst

• Telemetry Dashboard (Task 007a):

  • Mission Control UI for visualization and emergency stop.
  • Documentation: source/data/brain/dashboard.rst

• Serial Lexer & Parser (Task 004a/b):

  • UART input scanning (#...;;) and Float extraction (sscanf).
  • Documentation: source/data/embeddedplatform/serial_lexer_parser.rst

• Synthetic Data Generator (Task 008a):

  • Automated script in raven-sim to capture diverse training datasets.
  • Documentation: source/data/brain/synthetic_capture.rst

• Sign Recognition AI (Task 008b):

  • YOLOv8-based detection of Stop Signs and Parking Meters.
  • Documentation: source/data/brain/sign_recognition.rst

• Dead Man Switch (Task 004c):

  • Motor safety watchdog (500ms timeout) to prevent runaway vehicles.
  • Documentation: source/data/embeddedplatform/dead_man_switch.rst

Updated on Mon 02 Feb, 2026 by Hatem Soliman