Senior Design Projects

ECS193 A/B Winter & Spring 2019

Recyclable Material Automated Identification and Picking

Email **********
Name Michael Siminitus
Affiliation Owner - Waste Busters

Project's details

Project title Recyclable Material Automated Identification and Picking
Background Last year, Waste Busters sponsored a collaborative project with teams from the Mechanical Engineering department and the Computer Science department to develop a Computer Vision based system for identifying and picking recyclables off of a moving belt using a webcam and AI that controller a Ragnar pick and place robot. While this project was a success, some further functionality and improved modeling would increase the usefulness of the system. We propose a similar structure as last year, with coordinated teams from each department working on respective aspects of the system to bring it to a useful and functional level of performance.
Description While basically functional for identifying cups on a belt, we would like the teams to improve the system to enable it to pick up more than just cups in the waste stream. The Mechanical Engineering team is tasked to improve the placement and presentation of materials on the belt, spreading the material out for easier identification and picking. The Computer Science team would have the goals of improving the model and training to include recognition of common recyclable items such as aluminum cans, PETE plastic bottles, etc… Incorporating a nearIR camera or other method of identifying plastic resin type and combining this data with CV output to classify plastic cups by type ( PLA, PETE, PS ). Additional goals include improve accuracy of identification, speed of operation, reduce # of missed items Continue making the GUI end-user friendly.
Deliverable Working system that can identify and correctly sort 3D recyclable materials A well trained model that has a low rate of incorrect identification A User friendly GUI
Skill set desirable Computer Vison, AI UX design
Phone number **********
Client time availability 30-60 min weekly or more
IP requirement Open source project
Attachment N/A
Selected No
Team members N/A