RoboConda

Full Design Report: https://docs.google.com/document/d/1n9LT0mu90xYzykV9PIwDzWL7QmbWmHaDTMj11oVxwj4/edit?usp=sharing

Introduction

Problem Statement

Contamination rates in some recycling programs can be as high as 25%. Contaminated waste, such non-recyclables and organics, can cause the machines to jam or break down. A single contaminated material could send an entire batch of recyclables to the landfill.

Target Consumers & Their Challenges

The vast majority of Materials Recovery Facilities (MRFs) still use humans to sort through tons of waste and sort out contaminants. Not only is it a dehumanizing and dangerous job, it’s also inefficient. The RoboConda is designed to be able to be utilized for MRFs of all shapes and sizes.

Solution Overview

My Solution

RoboConda (courtesy of ChatGPT) uses a recycled Rubik’s Snake inspired robotic arm combined with artificial intelligence image recognition to remove contaminants from MRF conveyor belts before it can disrupt operations.

How It Works

RoboConda will utilize BakeryScan, an image recognition artificial intelligence originally designed to identify various baked goods and is now being used to detect cancerous cells, in order to identify non-recyclable and organic contaminants.

After identifying which objects need to be removed, the robot will use a system of rotating triangular prisms, inspired by the Rubik’s snake, in order to position itself above the contaminant.

From there, a suctioning vacuum tube will activate to remove the contaminant from the conveyor belt and place it into the appropriate receptacle. Organics will be set aside for compost, specific plastics will be set aside in order to be recycled into more RoboCondas, and non-recyclables will be removed in order to prevent damage to the system.

Technical Features

Nearly every component of the RoboConda is modular and adjustable for conveyor belts of all shapes and sizes. Not only that, but every fastener utilizes a standard 3/8”-16 bolt to ensure easy repair and replacement.

The mount between the snake arm and overarching beam contains space for any electronics that will be necessary to control the arm’s movements and vacuum controls.

The design was also entirely modeled in OnShape. A completely free in-browser CAD software. I want to show other aspiring engineers that you don’t need expensive software to start creating professional designs.

CAD Model

Video of exploded view of parts and configurations: https://drive.google.com/file/d/1ALhgYTnQtfMz3jp5GsB3ybeaxj9j4BCk/view?usp=sharing

Snake Arm

How it works

The snake arm’s deceptively simple looks contribute to both its modularity and extremely precise movement. Constructed of segments containing no more than four contact surfaces, the snake arm is designed for modularity and repairability. All of the vulnerable electronics are located outside of any moving pieces in a box above the system.

Each of the contacts in each component of the snake arm communicate with each to determine the current orientation of the robot. Each possible configuration is translated into a series of binary outputs. Each position on the conveyor belt will be assigned the appropriate binary sequence, allowing the robot to quickly and efficiently determine where it needs to move.

This method of operation was inspired by the Configuration analysis of a reconfigurable Rubik's snake robot Research paper which can be found at https://www.researchgate.net/publication/328126552_Configuration_analysis_of_a_reconfigurable_Rubik's_snake_robot

User Story/Walkthrough

User Experience

The MRF would receive a kit with everything needed to set up the RoboConda. The two conveyor belt mounts are placed onto the sides of the belt and screwed and secured into place with ⅜”-16 bolts and nuts. The top arm is adjusted to the appropriate size and secured together with more bolts. Next, the robots and all their components will be installed before the computer vision system is turned on. It will run some preliminary checks to ensure it recognizes the correct items. Finally, the system will be operable.

Sustainability

Not only does the RoboConda directly contribute to ensuring the smooth operation of sustainable recycling operations, it is also sustainable in and of itself. Many plastics that are thrown away are able to be repurposed into filament for 3D printing. Inspired by the PRUSA 3D printer company who uses their printers to print parts for new printers, the RoboConda will collect suitable plastics to be repurposed into more snake parts.

Iterations and Improvements

Major Iterations

The idea began utilizing a standard 4/6-axis robotic arm, however with an average cost between $25,000 and $400,000, this did not fit with the accessibility I had in mind. Next, I experimented with the idea of a gantry robot, however these are also slightly out of budget and lack the mobility I desired. Finally, I got inspired by the Rubik’s snake and landed on my final design.

During the mount design process, I fully completed a mount for a 3’ conveyor belt before I remembered that’s not the only size. I proceeded to redesign everything from the ground up with modularity and adjustability in mind to account for communities of all sizes.

Next Steps & Areas of Improvement

The very first step would be installing the overhead cameras in various MRFs in order to begin training the AI on what contaminants look like. I would also like to make the mount to the conveyor belts adjustable to accommodate different sized belt walls. Implementation would begin with smaller local MRFs to identify what needs to be improved and to instill confidence in the larger facilities. One major improvement that needs to be made is figuring out a solution for areas with poor connectivity as the camera would likely need it to communicate with the AI.

In terms of the presentation, with more time I’d like to import the proper electronic components into the CAD file, but as it currently stands, there’s not enough time for me to familiarize myself with electrical engineering enough to feel confident in my selections.

Built With

• onshape