Unity Simulation Sample Project: Warehouse Robot
Unity Technologies
$0.0
(no ratings)
Jump AssetStore
Warehouse Robot is a 3D sample project that demonstrates the usage of Unity for environment generation, Perception SDK for RGB images and semantic segmentation, and Unity Simulation for running it at scale to generate a dataset that can be used for training a machine learning model. This was created with Unity 2019.4 and uses the Universal Render Pipeline.
The project includes a SimulationController, a GameObject in the hierarchy that is responsible for driving the simulation; Robot NavMeshAgents; the Perception Package camera and custom logging tools for enabling RGB capture, semantic segmentation, and bounding box annotations; and the Simulation Client Package, which exposes C# APIs to automate the USimCLI workflow and enables menu options for building, generating app parameters, and executing a USim run.
Further details about the project and exposed parameters are available in the Documentation folder.
How to extend this project
This project can be extended to simulate robot navigation in differing environments. The robot's navigation system can be replaced with a custom model, trained on the synthetic dataset generated in the USim runs.
Further documentation on the Perception SDK can be found here.
You can learn more about Unity Simulation at our website, and get started by visiting our Unity Simulation forum.
Note: builds for Unity Simulation currently must target the Linux platform. This package will come with an example scene that requires certain project settings in order to function correctly, which will overwrite existing settings. Import this package into an empty project for best results.
For further troubleshooting, please email us at simulation-help@unity3d.com.
