1. system design Final report
C-4P5
Automatic Beer Bottle Filler and Capper
Aaron Landy
IMDL Fall 2013
November 21, 2013

2. Design and Operation

3. Overview
System Block Diagram

4. Overview

5. Software Design Three primary systems: Raspberry Pi
Master system control
Image processing
Slave management
Software distribution
Horizontal Axis Arduino
Primary real-time control
Bottle searching
Bottle handling
Vertical Axis Arduino
Filling control
Cap retrieval
Capper control
Height Adjustment

6. Raspberry pi C++, OpenCV Python Master system control Image processing
User interface
Progress monitoring
State save, resume
Testing control
Image processing
Bottle volume estimation
Bottle height estimation
Slave management
Master <-> Slave serial communication
Software distribution
Build in place
Arduino Compile, Upload
C++, OpenCV
Python
make, g++, SCP

7. Horizontal Axis Controller
Bottle searching
CDs to ADC input
Digital threshold monitoring
Trigger Claw on detect
Primary real-time control
Task scheduling
Sensor reporting
Motor Control
Background state machine
Sensors
Periodic background sensor polling
Real-time digital filtering
Serial communication
Program execution
Arduino

8. Questions ?

9. Motivation Homebrewing is fun, bottling beer is hard
Time consuming process requires individually filling and capping bottles of different shapes and sizes
~50 bottles per 5 gal batch

10. Bottle Detector Approaches
Simple Filtering and Area
Blob Detection
Inverse Binary thresholding to isolate bottle against background, filter noise
Count pixels for bottle area
Measure longest range of pixels for bottle height
OpenCV SimpleBlobDetector:
Successive binary thresholding
Find contours in binary images to extract connected components
Group keypoint centers across binary images
Estimate blob centers, radii from grouped components

11. Threshold Calibration
Source 24 32 40 48 64 96
12oz
22oz

12. Pixel area Conversion
Area in pixels must be converted to real 2D area measure
Requires calibration calibration based on bottle sizes
Must factor in relative positions of bottle and moving camera
Area
Width
20587 90
22468 96
29708
135

13. Objectives Fill bottles with precise amounts
Cap bottles securely without breaking
Support inconsistent bottle shapes and sizes
Seek/retrieve bottles automatically
Do not contaminate, oxidize, damage, or waste beer

14. Mechanisms Four primary mechanisms: Bottle searching/retrieving claw
Infrared and bump sensors to locate bottle
Filling tube
Stainless steel tube lowered into bottle
Tube opening must be pressed against bottom of bottle to minimize agitation to prevent oxidation or contamination
Cap retrieving/placing magnet
Caps must be picked up from top of bottle and held during filling
Cap must be replaced precisely on top of bottle
Capping actuator
Forceful linear actuator with capping “bell” crimps/seals cap onto bottle

15. Electronics Raspberry Pi (Model B rev. 2)
Infrared Range Finder and bump to detect bottles
Raspberry Pi Camera module to calculate bottle height and volume
Arduino + Stepper Motors + Motor Driver to drive linear motion axes
Servos to open and close, turn claw