simpleVO

This repository is an example for using g2o to run visual odometry on KITTI dataset in C++

Installation Dependencies

1. pangolin: https://github.com/stevenlovegrove/Pangolin
2. opencv4 (build from source is preferred)
3. Eigen (sudo apt install libeigen3-dev)
4. g2o: https://github.com/RainerKuemmerle/g2o

Compiling the Code

mkdir build 
cd build
cmake ..
make

Running the Code

./run_kitti_stereo

Make sure you change the default.yaml file in config folder!

Visualizing the Odometry

./viewPose

The odometry data will be written to pose.txt file in result directory

Things to do

• Use ORB features
• Add loop closure thread with Bag of Words(DBoW3)
• Visualize the detected sparse featurepoints in the Pangolin Viewer