Lsd-slam Deep Learning (1)-Basic introduction and installation under Ros

Objective

Drawing on the two methods from rgb-d data processing-based on features and overall-based, the same can be considered for the entire picture matching, rather than just consider the feature points of ...

In general, this dense method requires a great amount of computation,Dtam: Dense tracking and mapping in real-time the use of the GPU to accelerate the operation of the method,semi-dense The Visual odometry for a monocular camera effectively reduces the computational time through a semi-dense filter, and even implements the algorithm in real time on a smartphone. Semi-dense Visual odometry for AR on a Smartphone

Personal understanding of the wrong place also please enlighten, reprint please indicate source: http://www.cnblogs.com/hitcm/

Main introduction

The main reference to a few articles, are from Tum. In the future, the algorithm and code are based on these articles, it is recommended to study carefully, must benefit.

semi-dense Visual odometry for a monocular Camera

semi-dense visual odometry for AR on a Smartphone

Lsd-slam: Large-scale Direct monocular SLAM

On the basis of the above articles, there is a complete lsd-slam algorithm.

The homepage of the algorithm is as follows

Https://github.com/tum-vision/lsd_slam

Http://vision.in.tum.de/research/vslam/lsdslam?redirect=1

Installation under Ros

Bo Master's programming environment is Ubuntu14.04+ros Indigo, in order to facilitate the record, the use of a virtual machine environment, may be a bit card. For the basic knowledge of ROS, please learn it yourself and don't repeat it here.

Install the Dependencies first

  sudo apt-get install ros-indigo-libg2o ros-indigo-cv-bridge liblapack-dev libblas-dev freeglut3-dev libqglviewer-dev libsuitesparse-dev libx11-dev     然后git下面的内容

   git clone https://github.com/tum-vision/lsd_slam.git lsd_slam      最后是rosmake

  rosmake lsd_slam

Run Lsd-slam

An example from the official, using the dataset below, 400+m

Http://vmcremers8.informatik.tu-muenchen.de/lsd/LSD_room.bag.zip

Unzip the

Then run the following 3 commands to see the effect

Rosrun Lsd_slam_viewer Viewer

Rosrun Lsd_slam_core Live_slam Image:=/image_raw camera_info:=/camera_info

Rosbag play./lsd_room.bag

Also can online another brother's article, link below

Http://www.bubuko.com/infodetail-1010849.html

Http://www.mamicode.com/info-detail-957983.html

Lsd-slam Deep Learning (1)-Basic introduction and installation under Ros