SLAM
Jump to navigation
Jump to search
Contents
SLAM
SLAM stands for Simultaneous Localization and Mapping
What this means
A system that builds a map of an unknown environment while simultaneously navigating the environment using the map. It can also track the previous positions of objects in order to determine its location within the space
What we're going to do
We are in the process of making a SLAM system for the bike in order to provide the navigation team with the required data for movement.
Our Current Planned Process
Zed ⇒ depth img ⇒ Laserscan ⇒ Occupancy grid
- Zed = the camera and software that comes with it. We are using the first Zed camera.
https://www.stereolabs.com/docs/getting-started/
- Depth image = An image that has various distances information. Zed already can create the depth image. Because Zed has two cameras, it can calculate the distance information knowing the distance between its own two cameras.
(We may eventually get a LIDAR that will allow us to skip these steps and directly have a laser scan)
- Laserscan = Uses the depth image to determine the shape of the objects. We plan to use the ROS package depthimage_to_laserscan to get this
http://wiki.ros.org/depthimage_to_laserscan
- Occupancy Grid = The data from the laserscan is transformed into an occupancy grid, which is essentially an array of data. We are planning on using the ROS package rtabmap for this step. We could also use gmapping, but rtabmap has better documentation. For rtabmap, we plan on using the published topics grid_map.From the ROS website, grid_map creates an occupancy grid generated with laser scans. It uses parameters with prefixes map_ and grid_. We will give this to the navigation team.
http://wiki.ros.org/rtabmap_ros
