Zero Ones Simulated

A simple simulator for a robot arena inspired by the Robot Uprising Micro Invaders competition. If you're interested in training a machine learning model on a similiar arena, take a look at the official Micro Invaders AI Simulator.

Quickstart

• You can use one of the pre built binaries (Windows, MacOS, Linux)
• Alternatively you can build and run the Unity project
  • Double click the MainScene from the editor (assets/scenes/MainScene)
  • The built version expects a configuration.json file and a folder called Images in the same folder as the binary. You can copy them to the build folder from the root of the project.
• Once the simulator is running, you can get the overhead video feed (in MJPEG format) from: http://localhost:8080
  • On some systems the standalone version can't output video while the Unity project is open

• You can control one of the robots with your keyboard (this is enabled in the default configuration for testing purposes)
• You can control another robot by sending semicolon delimited string of motor values as UDP packets to localhost port 3001 or 3002. The values shoud be between 100 and -100
• You can change the starting positions and control methods by editing the configuration.json
• The camera is slightly tilted and the flickering option for one of the robots has been enabled in the default configuration to help you develop robust solutions. In the real world, the camera is never perfectly aligned and aruco marker or ball detection will occasionally fail.

To test out controlling the robot via UDP you can for example use the following command (Windows does not have netcat by default)

echo -n '100;-100' | nc -u 127.0.0.1 3002

Or in Python:

import socket
sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
sock.sendto(bytes("100;-100", "utf-8"), ("127.0.0.1", 3001))

Resetting the simulation

You can reset the simulation by pressing the q key or by sending the command reset via UDP to the controlPort defined in the configuration (3000 by default). For example in Python you could do this:

import socket
sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
sock.sendto(bytes("reset", "utf-8"), ("127.0.0.1", 3000))

Manipulating object locations

You can drag and drop objects around. Left mouse button drags the object and right mouse button points the highlighted object towards the cursor.

Saving and loading

You can save the current object positions in configuration.json file by pressing ctrl + s or cmd + s. The previous configuration will be overridden. You can load saved positions from configuration.json by pressing ctrl + o or cmd + o. When running from the Unity editor you can press ctrl + shif + o instead to prevent the editor from stealing the keypress.

On MacOS when using the pre built binaries App Translocation may prevent loading or saving the configuration. This is also why the configuration file was moved inside the .app folder. You can still edit it by right clicking the .app and selecting Show Package Contents.

Reading the video stream

The video stream can be read by any client that supports MJPEG. For example in Python you could do this with OpenCV:

import cv2
capture = cv2.VideoCapture("http://localhost:8080")
ret, frame = capture.read()

Aruco markers are placed on the robots and at the corners of the arena. This way you can correct for camera misalignment or compensate for perspective if the camera is not directly overhead. The arena corner markers have the following IDs and layout:

• Top: left: 46, right: 47
• Bottom: left: 48, right: 49

Note that version 0.0.6 used partially different IDs in a different order.

A python example of how to apply perspective transformations based on aruco markers at the arena corners can be found here.

Configuration

Once you've built the project, you can use the configuration.json file to change the configuration without the need to rebuild the project. All the units are metric and the coordinate system is left-handed (Unity uses a left handed coordinate system). On the pre-built MacOS version the configuration file and images folder is inside Zero-Ones-Simulated.app. You can access them by right clicking the app and selecting "Show Package Contents".

Option	Type	Notes
quality	integer	The Unity graphics quality level
timeScale	float	Can be set to e.g. 1.5 if you want to simulate faster than real time
controlPort	integer	This port will listen to commands such as reset to reset the simulation
streamFPS	integer	The target FPS for the video stream. On a slow system you probably want to go lower than the default 25. The process of capturing and encoding frames is quite slow at the moment so trying to get over 30 might not be a good idea even on faster systems. The FPS limit is not set very precisely. In certain situations you may get a higher FPS than the set value.
streamResolution	integer	The resolution of the simulation stream (and window). The same resolution will be used for width and height (Can be omitted if streamWidth and streamHeight are set, which we recommend so you can test your solution works with non-square aspect ratios)
streamWidth	integer	The width of the simulation stream (and window) in pixels. (Optional)
streamHeight	integer	The height of the simulation stream (and window) in pixels. (Optional)
streamPort	integer	The port for the video stream
cameraOffset	array	A list of six floating point numbers in the following order: x offset, y offset, z offset, x angle, y angle and z angle in degrees. This affects the overhead stream camera and can be used to test camera misalignment. An example of a slightly misaligned camera: [0.13, 0, 0.02, -3.4, 1.3, 0] or camera at the side of the arena: [1, 0, 0, -27, 0, 0]. A perfectly aligned camera would use zero offset values: [0, 0, 0, 0, 0, 0]
robots	array	A list of robot objects
dynamicObjects	array	A list of balls or obstacles

Robot object

Robot objects have the following properties

Option	Type	Notes
marker	string	Relative path to an image file that will be displayed on top of the robot
flickerProbability	float	The probability of the marker being hidden at a randomly selected time. Between 0 and 1. For example 0.0005 is a good starting value. This can be used to make sure tracking a robot works even if the marker is sometimes blocked by e.g. a reflection. (Optional)
flickerDuration	float	Duration of hiding the marker in seconds (Optional)
color	string	A hexadecimal color code
control	string	One of the following: arrows, wasd, udp:3001 (the port can be any available port), random (useful for testing against moving opponents)
position	array	A list of three floating point numbers and an optional integer in the following order: x, y, z, angle in degrees

Dynamic object

Dynamic objects have the following properties

Option	Type	Notes
type	string	One of the following: ball, flickering-ball, ghost-ball, cube, traffic-cone
color	string	A hexadecimal color code
position	array	A list of three floating point numbers and an optional integer in the following order: x, y, z, angle in degrees
mass	float	Mass in kg
size	float	The size of the object in meters
value	integer	The value given for delivering the object to a goal
delay	integer	How many seconds to delay releasing the object to the arena (Optional)

Debugging

You can find the logs that the simulator outputs here:

• Mac: ~/Library/Logs/Zero Oness Given/Zero Ones Simulated/Player.log
• Linux: ~/.config/unity3d/Zero Oness Given/Zero Ones Simulated/Player.log
• Windows: C:\Users\username\AppData\LocalLow\Zero Oness Given\Zero Ones Simulated\Player.log

Running a downloaded binary

You may see warnings of the binary being from an unidentified source. This is expected since we have not signed it. On MacOs something called App Translocation may also occur, where the application file is moved to a random location before execution as a safety precaution. This may break loading and saving the configuration.