Visual SLAM

Requirements

1. ROS (tested with kinetic)

2. sudo pip install kitti2bag (for conversion to rosbag)

Run the process

1. Clone the repo to your workspace https://github.com/srikanthmalla/visual_slam.git

2. cd visual_slam (go in to the package)

3. Download Kitti Visual Odometry Dataset sh download_dataset.sh (automatically into datasets folder)

4. Compile the code using catkin_make

5. Run roslaunch visual_slam visual_slam.launch

RunTime

Input data (camera) is 9.690 hz

With cpu_flow (KLT)~ program runs at 3hz

For using gpu_flow from openCV, please check setup_notes.txt

To remember

• In opencv z is considered forward, but x is considered backward in ROS (kitti) and z down.