stereo_lidar_recorder

Realtime Stereo Visual Odometry

StereoLiDARRecorder Build Instructions

1. Overview

This guide provides step-by-step instructions to build the StereoLiDARRecorder project for:

• Raspberry Pi 4 using Docker for cross-compilation
• PC (x86_64) for native development

2. Prerequisites

Ensure you have the following installed on your PC:

• Docker (for Raspberry Pi 4 cross-compilation)
• CMake (>=3.10)
• GCC/G++ (for native builds)
• OpenCV & PCL (for both target platforms)

3. Building for PC (x86_64)

Step 1: Install Dependencies

Run the following command to install required dependencies:

sudo apt update && sudo apt install -y cmake g++ libopencv-dev libpcl-dev

Step 2: Build the Project

Navigate to the project root and create a build directory:

cd ~/stereo_lidar_recorder
mkdir -p build_pc && cd build_pc

Run CMake and compile:

cmake .. -DCMAKE_BUILD_TYPE=Release -DCMAKE_TOOLCHAIN_FILE=toolchain_pc.cmake
make -j$(nproc)

Step 3: Run the Executable

./StereoLiDARRecorder

---

4. Cross-Compiling for Raspberry Pi 4 via Docker

Step 1: Build the Docker Image

Inside the rpi-cross-compile folder, create a Dockerfile if it doesn’t exist:

FROM ubuntu:20.04
RUN apt update && apt install -y cmake g++ aarch64-linux-gnu-g++ libopencv-dev libpcl-dev
WORKDIR /workspace

Now, build the Docker image:

cd ~/stereo_lidar_recorder/rpi-cross-compile
docker build -t rpi-cross-compile .

Step 2: Run the Docker Container

Run the container and mount the workspace:

docker run -it --rm -v ~/stereo_lidar_recorder:/workspace rpi-cross-compile

Step 3: Build the Project

Inside the container:

cd /workspace
mkdir -p build_rpi && cd build_rpi
cmake .. -DCMAKE_TOOLCHAIN_FILE=/workspace/toolchain_rpi.cmake -DCMAKE_BUILD_TYPE=Release
make -j$(nproc)

Step 4: Transfer and Run on Raspberry Pi

Copy the binary to the Raspberry Pi:

scp StereoLiDARRecorder pi@raspberrypi:/home/pi/

On the Raspberry Pi, run:

chmod +x StereoLiDARRecorder
./StereoLiDARRecorder

---

5. Troubleshooting

1. CMake Cannot Find OpenCV or PCL

Make sure OpenCV and PCL are installed. If needed, specify their paths in CMakeLists.txt:

set(OpenCV_DIR "/usr/lib/cmake/opencv4")
set(PCL_DIR "/usr/lib/cmake/pcl")
find_package(OpenCV REQUIRED)
find_package(PCL REQUIRED COMPONENTS common io)

2. Docker Image Not Found

Rebuild the Docker image:

cd ~/stereo_lidar_recorder/rpi-cross-compile
docker build -t rpi-cross-compile .

3. Binary Not Executable on Raspberry Pi

Ensure the binary is built for aarch64:

file StereoLiDARRecorder

If incorrect, check your toolchain_rpi.cmake file for errors.

---

6. Notes

• Adjust the toolchain files (toolchain_pc.cmake and toolchain_rpi.cmake) as needed.
• Use make clean before rebuilding if errors occur.

---

7. License

This project is licensed under the MIT License.

---

8. Author

Batchaya Yacynte

stereo_lidar_recorder

Realtime Stereo Visual Odometry

This repository implements a stereo visual-LiDAR odometry algorithm designed to estimate the motion of a robot or drone using data from stereo cameras and LiDAR sensors. The project aims to integrate depth information from both visual and LiDAR sources for robust and accurate localization.

Currently, the visual odometry component is fully functional, using stereo camera data to compute relative pose (rotation and translation). The integration of LiDAR data is a planned enhancement to improve accuracy and robustness in environments with limited visual features.

Key Features: Stereo Camera-Based Odometry:

Processes left and right camera images to estimate depth via triangulation. Computes motion using feature-based techniques for 3D point reconstruction. Future LiDAR Integration:

Plans to incorporate LiDAR data for enhanced depth estimation and odometry performance, especially in visually challenging environments. Real-Time Focus: Optimized for lightweight computation, with potential for integration into drones or autonomous robots.

Applications: Autonomous navigation for drones and mobile robots. 3D mapping and reconstruction for terrain and environment studies. Multi-sensor fusion for robust odometry and localization. Tools and Technologies: Languages: Python. Libraries: OpenCV, NumPy, Open3D (planned for LiDAR integration). Data: Supports stereo images and calibration parameters; LiDAR data integration in progress. How to Use: Clone this repository. Provide stereo camera images and calibration parameters. Run the visual odometry algorithm to estimate the camera's trajectory. [Future] Integrate LiDAR data for improved accuracy. Contributions are welcome! Feel free to explore and help enhance the LiDAR integration or optimize the visual odometry component.