openFrameworks addon that implements SLAM with a Lidar
README
ofxLidarSlam
openFrameworks addon for performing SLAM using Lidar data.
Uses the lidar-slam library
Currently only implemented for Ouster Lidars, but any other could be used if the data is "adapted" properly.
Only tested on Macos BigSur on an Intel Mac.
Requirements
Use a nightly build of openFrameworks. It will not compile on the current release (0.11.2) ofxOusterLidar. ofxDropdown ofxCeresLib ofxPCL ofxGuiTabs
Make sure you have g2o installed.
brew install g2oImportant
M1
On M1 macs, brew installs into a different directory, so you will need to modify the addon_config.mk file and put into it the correct file paths for the librariesBoost
Boost library is partially used within OF and it is included with it, but some other parts of it are needed so it is also installed using brew and wee need to link to those libraries. You will need to remove from your project the paths to Boost included in OF. On Xcode, go to the project settings and scroll all the way down and leave empty the following fields:LIB_BOOST_FS LIB_BOOST_SYSTEM
- HEADER_BOOST
Workflow
This addon is able to perform slam either in realtime or offline. Realtime only depends on having a fast enough computer. For offline currently you need to record using the OusterStudio app and then open the saved files in this app.
ofxAddons