Created by Jani, Abhee, last modified by Lai, Charles on Dec 12, 2021

Name

NetID

Section

Abhee Jani

abheej2

ECE 110 

Jia-Ye, Lai

jiayeyl2

ECE 110 

Statement of Purpose

Sortbot is a robot that will be created for the purpose of breaking down large piles of objects into organized sections specified by the user. SortBot will operate on wheels and chassis and have a clamp attached to the front capable of picking up and dropping off these objects. Through visual recognition, SortBot will be able to operate on a certain path designated for the category the object picked up belongs to. After each iteration of sorting, SortBot will be able to return to the user and pick up the next object without the need of intervention, allowing for there to be a seamless sorting process and ease of use for the user. 

Background Research

Some inspirations for our project were autonomous bots that are used for cleaning and organizing around households in general. This includes products like the Roomba or the Shark Robotic Vacuum. In addition, we did some research on robotic parts used in factories for industrial usage that are able to discern from large varieties of objects to sort them accordingly. We wanted to give this kind of industrial technology to the typical household. This kind of usage was something that people should be able to find useful in their everyday lives. The practicality of the robot was vital to our vision of the project. 

 

Block Diagram / Flow Chart 

                         Initial Design

 

 

                           Revised Design

   

                         Thought Process

 

System Overview

              Video Link: https://www.youtube.com/watch?v=M6XTartb0Kw

              Final Paper Link: https://docs.google.com/document/d/1qyofohGkxUB_9HtA28nb6yfVxR4Zzu_mlwfTO0WPbWQ/edit?usp=sharing

Parts

Index

Item

quantity

Price

Part Link

1

Camera Module

(913-2673)

1

25

https://www.digikey.com/en/products/detail/raspberry-pi/913-2673/6152811

2

Raspberry Pi 3 Model B+

1

25

Honor Lab Inventory

3

Tank Chassis

1

25

https://www.digikey.com/en/products/detail/dfrobot/ROB0153/10230090

4

Mini-SerVo Motors*3

2

3.62

https://www.digikey.com/en/products/detail/dfrobot/SER0006/7597224

5

Robotic Arm




6

QUAD AA BATTERY HOLDER

2

2.19*2

ECE Supply Center

7

UBEC

(1385)

1

9.95

https://www.adafruit.com/product/1385?gclid=CjwKCAjwhaaKBhBcEiwA8acsHBSX5xg3NP1VpU_8gDoJsuphd8Ps_11S9R6vIUDdgEINRrvtdM4LihoCRmAQAvD_BwE

8

AA battery

7

0.47*7

ECE Supply Center

9

ADATA 16GB microSDHC UHS-1 Card

1

4

ECE Honor Lab Inventory

Sum

Total Spending

N/A

103.86

N/A

Possible Challenges

  • Programming a robot that can come back to a certain location can be challenging.

  • The processing ability of the Raspberry Pi might not be enough to provide an accurate result for image identification.

  • We have to figure out what mechanism to implement for capturing the designated object

References


Attachments:

zen_wireframe.png (image/png)

Comments:

Quick heads up: you can't buy parts from eBay this semester. You'll need to change your parts list. 

I'm having a little trouble understanding what your sorter would look like. I'm imagining a tank-like robot with a claw on top. Please clarify what project you are proposing so we can go into further detail with our feedback.

Assuming this is the case, I'm not sure if image recognition running on an Arduino will be fast enough for your purposes. Consider using something like a raspberry pi instead, so you can use vision libraries. I also would suggest that you look into what components you'll need to put all these pieces together. I don't see anything power related in your budget, for example. 
Posted by as85 at Sep 21, 2021 18:57

Approved 
Posted by dbycul2 at Sep 23, 2021 18:59

Very cool project.


Some notes:
I highly recommend you initially limit the range of objects that the device can sort. It will drastically simplify many aspects.

For example, limit it to only sorting colored blocks into their respective colors, your mechanism to push the blocks can be simplified (since it's only one object you need to worry about)



  • The processing ability of the Raspberry Pi might not be enough to provide an accurate result for image identification.

It will probably be sufficient but could depend more on the thing(s) you are identifying (and more relevantly how large/complex the model needs to be). 
Posted by weustis2 at Sep 27, 2021 23:13

Group #15
Posted by dbycul2 at Sep 28, 2021 00:15