Indoor and outdoor Localization
using factor graph based GPS and 3D Lidar sensor fusion
Using GTSAM to perform sensor fusion using GPS like data and 3D LiDAR based localization
Description
This project was primarily an exploration into factor graphs and how we could do state estimation using them. But we had a few other objectives; to get a AMR robot running, and set up a 3D lidar to collect data , use open source packages to perform 3D localization and to use a variation of pure pursuit controller to follow a set of waypoints indoors accurately.
About
​
We made a prototype frame and mounted a Velodyne Puck 3D lidar on the Scout 2.0 AgileX platform, and then performed 3D SLAM and localization using open source ROS packages.
​
We imprlemented a pure pursuit controller and used a series of way points to define a path through an indoor and outdoor setting.
​
We finally moved the robot around to collect series of pose information from the lidar based data, where a part of the path was indoors and a part of it was outdoors.
Next we simulated some GPS data (the ones we got from the robot were inaccurate due to buildings and weather interference) and then used factor graphs to fuse the estimate to show that we can get good estimates for both outdoor and indoor with both the GPS data
​
The Team
Aadith Kumar