Rover Project

Robotics project for the final year at INSA Toulouse, featuring a modular system for hardware control, mapping, and vision-based autonomous/semi-autonomous navigation.

1. Overview

This repository serves as the main entry point for the project, which is structured into the following submodules:

• Rover-Pi: Handles hardware control (motors, sensors, camera).
• Rover-Coral: Manages higher-level logic and mapping functionality, leveraging the Google Coral Dev Board.
• Rover-Protos: Contains the gRPC proto definitions for communication between modules.
• Rover-Vision: Handles computer vision testing and model training.

Each submodule contains its own README with detailed setup and usage instructions.

2. Features

• Modular design with gRPC-based communication.
• Real-time mapping and path tracing.
• Remote control with hardware interfacing for sensors and motors.
• RTSP video streaming from the PiCamera2 on the Raspberry Pi Zero.
• Testing and training of computer vision models for resource detection.

3. Quick Start Guide

Cloning the Repository with Submodules

git clone --recurse-submodules git@github.com:bastiankrohg/smart-rover.git
cd smart-rover

If submodules are not cloned, initialize them:

git submodule update --init --recursive

Setting Up the Environment

1. Follow the setup instructions in the respective submodules:

   • Rover-Pi Setup: Configure and control the rover's hardware.
   • Rover-Coral Setup: Manage mapping, gRPC communication, and high-level control.
   • Rover-Vision Setup: Train and test computer vision models for resource detection.
   • Rover-Protos Setup: Update and compile gRPC protobufs.

2. Configure communication:

   • Assign static IPs for the Raspberry Pi and Coral Dev Board.
   • Ensure the devices can communicate over the same local network.

Running the System

1. Start Hardware Control:

   • Launch the rover-pi control script on the Raspberry Pi:

     python3 rover-pi/pi_v2.py

2. Start Mapping and Control:

   • Launch the rover-coral mapping and control module on the Coral Dev Board:

     python3 rover-coral/control.py

3. Access RTSP Stream:

   • View the live video stream from the PiCamera2 at:

     rtsp://<pi-ip>:5054/cam
     

   • Replace <pi-ip> with the actual IP address of the Raspberry Pi.

4. Interact with the Rover:

   • Use gRPC clients or manual scripts to control movement and mapping.

4. RTSP Streaming with Mediamtx

The Raspberry Pi Zero streams video from the PiCamera2 using Mediamtx.

• Configuration:

  • Resolution: 640x480
  • Bitrate: 500 kbps
  • Horizontal and vertical flip enabled for correct orientation.

• Access:

  • Stream Address: rtsp://<pi-ip>:5054/cam

Mediamtx is installed as a systemd service and starts automatically on boot.

5. Future Improvements

• Integration:

  • Synchronize mapping and hardware control for seamless operation.

• Autonomous Navigation:

  • Implement algorithms for obstacle detection, path planning, and search patterns.

• Telemetry Dashboard:

  • Develop a dashboard to display real-time rover telemetry and performance metrics.

• Enhanced Calibration:

  • Integrate odometry data from motor encoders for precise movement tracking.

• Vision-Based Navigation:

  • Incorporate trained models from rover-vision for real-time decision-making.

6. Archived Notes

References

• 4tronix Assembly Guide
• Raspberry Pi OS
• Programming M.A.R.S. Rover

SSH Configuration

• Default IPs:

  • Rover: 192.168.0.168
  • Rover2: 192.168.0.169

• Connect via:

  ssh rover@192.168.0.168

• Safe Shutdown

  sudo shutdown -h now

• Adding Known Wi-Fi Networks

  • Setting up multiple known wifi networks

• Troubleshooting

  • Missing python packages:

  pip install rpi-ws281x RPi.GPIO smbus

  • Missing I2C/SPI configuration: Enable in sudo raspi-config.