S.L.A.M and Tracking Technology
S.L.A.M stands for Simultaneous Localization and Mapping. SLAM and tracking technologies make AR better in every way. SLAM is a technology that is used in computer vision systems that gathers visual data from our world around us and incorporates that so that the machine can use that information. This is visual input. This technology builds a map in a sense. This map is created through varying sensors in the camera.
For all this to work S.L.A.M follows this algorithm below
The Parts of the System
Sensor Data: For phones, this is the camera, accelerometer, and gyroscope.
Front End: Short term feature extraction, these need to be related to actual landmarks. Basically, map points. Long term is what reduces drift and this is done by knowing what it has already seen and recognizing it.
Back End: This helps connect different frames.
Slam Estimate: The overall result. This includes tracked features and their locations and the camera in the real world.
Why SLAM is a Need?
Spaces that are already tracked and known are easy to navigate, but what about the unknown. SLAM is able to get through places with no previous GPS signals or maps existing. SLAM is best when there is not a reference point.
For self-driving cars, SLAM is what is behind them. With Google's current self-driving cars, SLAM uses a LIDAR sensor to create 3D maps of the current areas surrounding. In under 10 seconds, this is done. This technology is also featured on Mars. When certain landmarks need to be revisited, SLAM technology tracks them down and gets there.
Basic Tracking Technology Information
The 3 Important Terms of Tracking
Tracking, Calibration, and Registration
From the book AR Principles and Practice, it covers these.
Tracking: This describes the dynamic sensing and measuring of AR systems.
Calibration: Compares measurements from two different devices.
Registration: Alignment of coordinate systems between virtual objects and real objects.