Skip to content

Example Projects

Example projects are complete, ready-to-run projects, each built around a robot and a technique. Lucky Engine ships with one, Welcome, and the Content Vault holds more. They are the quickest way to see robot control, gripping, procedural content and data recording working together in a real scene.

In short

The bundled Welcome project opens the first time you start the editor: click Continue on the Welcome screen, then press Play to watch a Franka Panda arm keep a marble run going. Other examples, such as Piper Pattern Stacking and Walking Pick and Place, are listed under EXAMPLES on the Welcome screen and in the Content Vault's Examples category. Download creates a new project from one and opens it.

  • Welcome

    A Franka Emika Panda arm picks marbles out of a marble run and drops them back onto the track. Bundled with the editor.

  • Piper Pattern Stacking

    An AgileX Piper arm stacks blocks into a target pattern.

  • Walking Pick and Place

    A Unitree G1 humanoid walks to cups, picks them up, and moves them, combining a trained walking policy with IK.

Welcome

Welcome is the project bundled with the editor. The editor opens it the first time you start it, and whenever it can't find the project you used last. Its start scene is the Welcome scene.

The scene is a marble run. Four marbles roll down the track and come to rest in a catch at the bottom. A Franka Emika Panda arm, with Robot Controller and MuJoCo Scene components, watches the catch. Whenever a marble is waiting there, the arm hovers over it, reaches down, closes its gripper, carries the marble to a drop target and lets it go. The marble rolls down the run again, and the loop continues for as long as the scene plays.

To try it, click Continue on the Welcome screen and press Play. While the scene plays, press Enter to restart the pick-and-place routine.

The scene is set to the Path Traced render mode. On a computer that cannot path trace, the viewport uses Rasterized instead. See Rendering.

How it works

A script named Welcome, on the Panda entity, runs the routine as a sequence of timed steps. It never sets joint angles itself. Instead it sends the arm's motion graph a target position, a target orientation in degrees and a duration, then triggers the move, and the motion graph solves the joint motion with inverse kinematics. The gripper is a MuJoCo actuator that the script opens and closes directly.

Select the Panda in the Scene Hierarchy to see the script's fields in the Inspector, grouped under Targets, Timing and Grasp. Change the timings while the scene plays to see their effect. The Run Pick & Place button starts the routine, like Enter, and works only while the scene plays.

The track pieces are built from procedural graphs, each placed with a Procedural Mesh component. Select a piece to change its parameters in the Inspector, such as the length and drop of a straight. See Procedural Graphs.

Each completed pick-and-place also ends an Observer episode. If you record the scene, every cycle becomes its own episode in the dataset. Recording needs you to be signed in; see Recording.

For the script-level view of these pieces, see Working with components and Recording with the Observer.

Customise a copy

The editor can replace the bundled Welcome project with a fresh copy. Before you change it, choose File → Save Project As... to save your own copy, and work in that.

Piper Pattern Stacking

Piper Pattern Stacking is built around the AgileX Piper, a 6-DoF arm with a parallel gripper. The arm stacks blocks into a target pattern. The project comes with the Piper robot, its materials and a ready-to-run scene.

Walking Pick and Place

Walking Pick and Place is a Unitree G1 humanoid fitted with five-fingered hands. It walks to cups, picks them up, and moves them by combining two kinds of control at once: a trained walking policy and IK.

It is the example to study for policy-and-IK co-control on a humanoid. The general pattern, where a policy and an IK chain drive different joints at the same time and a joint-ownership mask keeps them from overwriting each other, is described in Controlling robots: policies and IK. The five-fingered hands make it the richest grasping example in the set.

Opening an example

Examples other than Welcome are downloaded from the Content Vault, so you need a network connection. Each download creates a new project from a fresh copy of the example. There are two places to start one:

  • On the Welcome screen, find the example under EXAMPLES and click Download. Tutorial, where an example has one, opens its tutorial, and Read More shows a long description in full.
  • In the editor, open View → Content Vault, set the category to Examples, and click the download button on the example's card.

Either way, the Create Example Project dialog opens:

  1. Check the Project Name. It starts as the example's name.
  2. Check the Location, or click ... to choose another folder. The default is your projects folder, described in Creating a project.
  3. Click Create. A progress bar with a percentage and a Cancel button shows the download. When it finishes, the editor opens the new project.

Downloading an example again creates another, separate project under a new name or location. Your earlier copy is left as it is.

The examples in the Content Vault can change between releases, and the Examples category is the current list. In the Content Vault, an example that needs a different editor version shows Requires and a version range instead of a download button. The Content Vault also lists Welcome as an installed example. It can't be removed.

Where to go next