Burger Bun Detection

Project Description

I interned at Creator, Inc over the summer of 2018. Creator operates a restaurant in San Francisco with robots that make made-to-order gourmet hamburgers. Part of this process requires distributing a selection of sauces onto buns on a conveyance system. My project was to integrate machine vision onto the saucing system in order to improve the accuracy of sauce distribution. This was an inter-disciplinary project, where I selected appropriate camera hardware, wrote prototype bun-detection software using OpenCV and integrated it with a BeagleBone Black micro-controller

Software Implementation

The bun-detection algorithm that I wrote used OpenCV's Canny Edge Detector and Hough Circle Transform to identify the 2D contours of buns from images taken from within the saucing system. Inherent parameters of the transform, such as circle radius, edge contrast strength and circularity were tuned to distinguish the buns from other features in the image. The software was able to robustly detect buns invariant of lighting conditions in the restaurant.

© 2019 by Adnan Jafferjee. 

A failed distribution

Sauce distribution pattern is off-center and has messily splattered all over the box