TrackingExpert+ introduces a set of tools for augmented reality (AR) application development. In particular, it provides for 6-degree-of-freedom (DoF) pose measurement and 3D visualization capabilities. The focus is on industrial AR applications for tasks such as assembly training and task-support, maintenance, and others.
The software allows a user to detect objects and track their 6-DoF pose using point cloud data acquired from an RGB-D camera. The underlying pose estimation approach is model-based: it requires a 3D model of an object of interest as input data. Underneath the hood, TrackingExpert+ relies on features descriptors and statistical pattern matching to perform its task. Its feature descriptor employs surface curvatures of an object and the relative curvature distribution around a point. This approach renders the software highly robust in the presence of noise, occlusion, and other obstacles since it can work a minimum visible surface.
TrackingExpert+ provides the following features:
- Point cloud generation from an RGB-D image.
- Normal vector estimation.
- Point cloud sampling.
- Object detection in point clouds.
- Object 6-DoF pose estimation.
- AR assembly training example application.
Additionally, a Unity plugin is under development, which allows a user to employ the HoloLens along with TrackingExpert+ for any AR application.
All code runs (tested) on Windows 10, Version 1803. It requires the following 3rd party tools:
The code was prepared with Visual Studio 2019. Note that this is only the software the development team uses. All components should also work with never software veersion.
Follow the link to the TrackingExpert+ Installation and User Manual. Please leave comments if you think information is missing, incorrect, or incomplete.