Skip to content

Latest commit

 

History

History
94 lines (66 loc) · 2.64 KB

README.md

File metadata and controls

94 lines (66 loc) · 2.64 KB
Raw

Kitchenator

  • Kitchenator is powered by KitchenNet

All software tested on MacOS Sierra Version 10.12.6

Overview

The state is recorded and available on a server launched by kitchennet.py The clients hook into this server and either update or check on the state by calling different HTTP methods KitchenNet proceeds through a number of different states during an ingredient cycle. The states are set by a State Manager Client.

Information is passed between clients by updates to the states held by KitchenNet. All data is passed as JSON

Gesture Client:

Updates the gesture state in KitchenNet Gesture JSON:

{"gesture": 1}

Arm Client:

Checks KitchenNet to see if it should be 'going' (and if so, where to), 'standby', or 'paused' The arm client's GET request returns JSON of the form: {"armGoalPose": [x, y, pitch, yaw], "armGoalState": 'standby/paused/going'}

Vision Client:

Checks KitchenNet for a desired target, if so updates the target arm pose

Voice Command Client:

This MATLAB client listens on a button press and performs a cross-correlation against the known ingredients. It assigns the closest match to the user's speech as the goal ingredient. https://github.com/mozilla/DeepSpeech#using-the-python-package

Visual User Interface:

Access at 127.0.0.1:12346/userInterface Access debug screen at 127.0.0.1:12346/uiDebug Future will include

Dependencies:

  • Python2
  • bottle
  • httplib (Python3 requires a switch to http.client)
  • json

Operation

In one terminal, run:

python KitchenNet/kitchennet.py NOTE: If there are errors relating to the 'Enum' class, go to KitchenNet/callHandlers/states.py and change 'WORKING' from True to False (or vice versa)

In separate terminals, run:

python theClients/manager.py python vision/k_webcam_shapes.py Note: You need the orange/black - green/black - purple/black color tags displayed for this client to begin transmitting data to the server

In separate MATLAB instances, run:

Myo/MyoGestureDetection.m

and

theClients/voiceClient

and

RobotArm/run_arm3

Locate the MATLAB window spawned by voice client, click on it, speak the ingredient ("nutmeg" or "flour"), and then sit back and watch the magic happen.

States that the arm client can receive:

Color Poses from GET /getAllPoses

{ {"purple": {'x': 0, 'y': 0, 'z': 0, 'yaw': 0} }, {"orange": {'x': 0, 'y': 0, 'z': 0, 'yaw': 0} }, {"blue": {'x': 0, 'y': 0, 'z': 0, 'yaw': 0} } }

Arm Goal States from GET /getArmGoals

{ {"armGoalState": "go"/"stop"/"dump"/"undump" }, {"gripperState": "open"/"close"/"same"}, {"armGoalColor": "purple"/"orange"/"blue"} }