SimpleSim

This environment is a simple simulation engine to view and control a standard VEX Clawbot and move it via motor torques, similar to the interface used on the Jetson Orin Nano. This is in BETA and must be run from the repository directory until a package can be created and some features added, however feel free to test as you like.

Sim details: three.js render, ammo/bullet physics

OSHW Support: Windows/Mac/Linux, does not require a GPU

Quickstart

To install the needed libraries, you will first need tkinter. Note that while it comes built in on Windows, Mac/Linux will require an additional install.

# Linux
sudo apt-get install python3-tk

# Mac
brew install python-tk

For all platforms, you can then do a standard pip install.

git clone https://github.com/timrobot/SimpleSim.git
cd SimpleSim
python3 -m pip install -r requirements.txt

Run an example program to control the robot. You can use W(↑) A(←) S(↓) D(→) to control the robot navigation and P(arm ↑) L(arm ↓) and O(claw →←) K(claw ←→) to control the arm and claws.

python3 main.py

Example code

A full example code to manually control the robot can be seen here:

from cortano import RealsenseCamera, VexV5

if __name__ == "__main__":
  camera = RealsenseCamera()
  robot = VexV5()

  while robot.running():
    color, depth = camera.read()
    sensors, battery = robot.read()

    keys = robot.controller.keys
    y = keys["w"] - keys["s"]
    x = keys["d"] - keys["a"]
    robot.motors[0] = (y + x) * 50
    robot.motors[9] = (y - x) * 50
    robot.motors[2] = (keys["p"] - keys["l"]) * 50
    robot.motors[7] = (keys["o"] - keys["k"]) * 50

Description

A cube is positioned randomly inside of a standard 144"x144" field, with a controllable VEX clawbot.

Action Space	Box(-1.0, 1.0, (10,), float32)
Observation Shape	(12,)
Observation High	[inf inf inf inf inf inf inf inf inf inf inf inf]
Observation Low	[-inf -inf -inf -inf -inf -inf -inf -inf -inf -inf -inf -inf]
Import	from cortano import VexV5, RealsenseCamera

Action Space

Some actions are left intentionally blank to reflect the VEX microcontroller's disconnected ports.

Num	Action	Control Min	Control Max	Unit
0	Angular velocity target of the left motor	-100	100	percent
1	❌
2	Angular velocity target of the claw	-100	100	percent
3	❌
4	❌
5	❌
6	❌
7	Angular velocity target of the arm	-100	100	percent
8	❌
9	Angular velocity target of the right motor	-100	100	percent

Observation Space

VexV5 Clawbot

sensors, battery = robot.read()

Num	Action	Min	Max	Unit
0	Left Motor ang position	-inf	inf	position (degrees)
1	Left Motor ang velocity	-inf	inf	velocity (degrees/second)
2	Left Motor torque	-inf	inf	Nm * 1e3
3	Right Motor ang position	-inf	inf	position (degrees)
4	Right Motor ang velocity	-inf	inf	velocity (degrees/se5ond)
5	Right Motor torque	-inf	inf	Nm * 1e3
6	Arm ang position	-inf	inf	position (degrees)
7	Arm ang velocity	-inf	inf	velocity (degrees/second)
8	Arm torque	-inf	inf	Nm * 1e3
9	Claw ang position	-inf	inf	position (degrees)
10	Claw ang velocity	-inf	inf	velocity (degrees/second)
11	Claw torque	-inf	inf	Nm * 1e3

RealsenseCamera

color, depth = camera.read()

Name	Description	Shape	MaxValue	dtype
color	BGR numpy array	(360, 640, 3)	255	uint8
depth	Depth(m) numpy array	(360, 640)	10000	uint16

Camera API

camera = RealsenseCamera()

Property	Description
camera.fx	focal length (x axis)
camera.fy	focal length (y axis)
camera.cx	principal center point (x axis)
camera.cy	principal center point (y axis)

Environmental Details

The default unit is in meters, with the Z axis pointing upwards.

Color and depth frames are gathered from the Realsense Camera, which tries to mimic a Realsense D415. Note that by default, the simulator will display these frames on the control window. To manually control the robot, you can call the robot.controller.keys[keyname:str] api. For example

robot = VexV5()
while robot.running():
  a_pressed = robot.controller.keys['a'] # 0 or 1

Note that rendering the color and depth frames take a significant amount of time from the simulator, so it will cause step times to increase as well. To turn this off, set render to False:

robot = VexV5(render=False)

Aprilcube Details

Each of the cube's 6 faces contains one of three apriltags - tag16h5 {12, 13, 14}. The cube is 2.56"x2.56"x2.56", however when detecting tags remember that apriltags detect only the inner (black) portion. That means that in reality the tag's size parameter should be 2.56 * 3/4.