Grantee Research Project Results
Automated Waste Sorting at the Point of Disposal
EPA Contract Number: 68HERC22C0020Title: Automated Waste Sorting at the Point of Disposal
Investigators: Yhap, Charles
Small Business: CleanRobotics, Inc
EPA Contact: Richards, April
Phase: I
Project Period: December 1, 2021 through May 31, 2022
Project Amount: $99,640
RFA: Small Business Innovation Research (SBIR) Phase I (2022) RFA Text | Recipients Lists
Research Category: Small Business Innovation Research (SBIR)
Description:
CleanRobotics seeks to improve the US recycling system by using robotic sorting, powered by object detection, artificial intelligence and machine learning (AI/ML) to sort up between 2-4 streams (recycling, compost, landfill, etc) at the point of collection. This will improve the collection and sortation of recyclables in the US, where total collection rates are 35% and successful sortation rates are less than 50% (worse than random chance).
CleanRobotics developed a robotic sorting system called TrashBot. TrashBot uses cloud storage and machine learning to determine which object users have dropped into the receptacle and the objects’ level of contamination.TrashBot then drops the item in the proper storage receptacle on the inside of the system. TrashBot identifies contaminated items and keeps them from entering the recycling stream.
Each TrashBot unit is deployed with a screen. CleanRobotics uses these screens to provide educational content, especially as it relates to contamination. When users throw away contaminated recyclables, the system separates the contaminated item into the landfill and informs the user why the item was contaminated. In small scale trials, this method has proven effective at decreasing the amount of contaminates in the recycling stream.
TrashBot collects data on every item users deposit. The system sends data to a centralized location where AI/ML systems determine patterns and identify trends in waste generation habits of TrashBot users. Cloud connectivity allows individual TrashBot units to learn from the global TrashBot fleet. This will increase the quality of the data we collect and drive AI/ML-based improvements to the TrashBot system over time. We analyze data collected in two ways: 1.) In the specific TrashBot deployment (i.e. the stadium, office building, airport, etc.) 2.) Across all TrashBot deployments. This data gives the CleanRobotics team and building managers insight into the kinds of items users are throwing away and trends in municipal solid waste generation. This improves AI/ML quality and allows us to tailor the messages we display on the TrashBot screen.
CleanRobotics has conducted pilots and small-scale deployments of the TrashBot system. Customers include Google, The Port Authority of New York and New Jersey, AEG, Dallas Fort Worth International Airport, Pittsburgh International Airport, UNC Charlotte, and several other companies. In one year, TrashBots: 1.) Sorted 30,000 items with 90% accuracy , 2.) Diverted 1,800 lbs of recyclables compared to 650 lbs with conventional bins, 3.) Saved companies from emitting five tons of C02.
Progress and Final Reports:
SBIR Phase II:
Robotic, Artificial intelligence (AI) Powered Trash System for Facility Sorting and Auditing Waste and Educating Transient PopulationsThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.