A complete end-to-end demonstration in which we collect training data in Unity and use that data to train a deep neural network to predict the pose of a cube. This model is then deployed in a simulated robotic pick-and-place task.
This PR updates the version of the perception package used in the demo project to 0.8.0 and introduces a few additional changes and fixes:
The ROS-TCP-Connector and URDF-Importer packages have been moved to submodules from being HTTPS github URLs in the project's manifest. This change will alleviate potential credentials issues that users can encounter when using the Unity Editor to download package repositories directly from URLs.
The PoseEstimationScenario now inherits from the PerceptionScenario class (introduced in perception 0.8.0)
The capture package version was manually updated to fix some warnings about missing meta files.
This PR updates the version of the perception package used in the demo project to 0.8.0 and introduces a few additional changes and fixes: