hySLAM is a hybrid SLAM/SfM system designed for mapping

HySLAM Overview

hySLAM is a hybrid SLAM/SfM system designed for mapping. The system is based on ORB-SLAM2 with some modifications and refactoring.

Raúl Mur-Artal and Juan D. Tardós. ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Cameras. IEEE Transactions on Robotics, vol. 33, no. 5, pp. 1255-1262, 2017.

ORB-SLAM2 original code: ORB-SLAM2_repo

Modifications:

1. Support for multiple cameras.
2. Addition of a Multi-map, recursive data structure: a recursive tree structure is used to handle sub-maps. Sub-maps can be optionally registered with their parent to make keyframes and map points accessible to the parent or the sub-map can be kept private
3. Trajectory tracking: Per-frame camera trajectories are explicitly recorded as SE(3) transformations relative to reference keyframes, whose positions are continuously updated via optimization
4. Extensive code refactoring including converting Tracking to a state-machine, conversion of Mapping to a job based, parallel module, and addition of a separate Feature Extraction thread.

Example Use

hySLAM was used as the basis for a dual-camera SLAM system to map visually repetitive ecosystems such as grasslands, where conventional Strucuture from Motion techniques are unreliable. The dual-camera SLAM system uses a forward-facing stereocamera to provide localization information while a downward facing high-resolution "documentation" camera is used to record the ecosystem. Conventional SLAM is used to analyze the forward-facing stereocamera video. The trajectory of the stereocamera is then used to guide localization and mapping from the documentation camera as illustrated in the figure below:

The dual-camera SLAM system allows reliable mapping of repetitive ecosystems as illustrated below: A: Accurately reconstructed campus lawn using dual camera SLAM. B: SfM failure due to visual aliasing (blue squares represent aligned images). The three lines of images (highlighted in red) should be parallel but instead converge on a single point in the reconstruction

Dependencies:

1. pangolin
2. DBoW2
3. OpenCV

Installation

1. in Thirdparty, compile and install g2o: cmake .. make -jX sudo make install sudo ldconfig
2. compile and install hyslam in main CMakeLists set opencv directory cmake .. make -jX sudo make install
3. build binary vocabulary: ./tools/bin_vocabulary