Skip to content

MindExMachina/robosketch-rnn

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

RoboSketchRNN

Let a robot continue your drawings using SketchRNN.

How it works

Three main pieces enable us to (1) draw, (2) predict the continuation of our sketch, and (3) tell the robot to draw them physically.

These three pieces are:

  • Machina Bridge · an app that connects to a robot and creates a WebSocket server to give the robot instructions
  • SketchRNN · a neural network that predicts how to continue a sketch from an input drawing, server as an HTTP service
  • p5js · a drawing application that requests predictions from the SketchRNN services and tells the robot to draw them on paper

Usage

  • Connect to the robot with Machina Bridge
    • Execute Machina-Bridge_v0.1.0/MachinaBridge.exe.
    • Choose the make of your robot (e.g., UR).
    • Write the local IP of your robot (e.g., 192.168.0.172).
    • Click Connect.
    • (The Machina Bridge app should be now connected to the robot.)
  • Start SketchRNN as an HTTP service
    • Go to the sketch-rnn folder.
    • Run npm install.
    • Run node http-server.js.
  • Open the p5js drawing app's index.html

p5js drawing app commands

Here are the keystrokes you can use to control the p5js drawing app.

  • p · request a prediction to SketchRNN
  • s · toggle prediction visibility
  • h · send robot home
  • r · tell the robot to draw the latest doodle
  • c · clean existing drawing

Troubleshooting

Errors with node-gyp while trying to npm install sketch-rnn

  • node-gyp only works with Python 2.7, make sure this is the version you have installed and the one referenced by npm.
  • Install Visual C++ redistributable tools for Visual Studio 2015. I fixed it by going to the Modify and Repair panel of Visual Studio 2015 (as if you were to uninstall it) and clicking on Modify, then selecting to install all the C++ tools.