paper-plane envelope home office pencil quill pen image images camera play bullhorn connection mic file-text2 file-picture file-music file-play file-video copy folder folder-open folder-plus folder-minus folder-download folder-upload price-tag price-tags ticket phone envelop pushpin location compass map map2 clock alarm fax mobile bubble bubbles user users user-plus user-minus user-check quotes-left quotes-right search pie-chart stats-dots stats-bars airplane cloud-download cloud-upload earth link flag eye eye-blocked arrow-up-left arrow-up arrow-up-right arrow-right arrow-down-right arrow-down arrow-down-left arrow-left2 share amazon google-plus google-drive facebook instagram twitter rss youtube flickr dropbox linkedin file-pdf file-openoffice file-word file-excel

Mechanical Engineering Group Design Projects
UCL Mechanical Engineering


UCL Autosort Team 2020

UCL Autosort Team 2020

Team Members:

  • Patricia Achaerandio
  • Pierre Chidiac
  • Julia Elkouby
  • George Imafidon
  • Costas Paschalides
  • Ahmed Salem
  • Josiah Youd


Professor Mark Miodownik

The Project

Autosort’s project was the design and prototype of an automatic waste-sorting bin which identified item types using a multi-sensor sensing system that included computer vision, sound recognition and a decision-making artificial intelligence algorithm. The aim was to reduce contamination and cross-contamination of the present waste streams to increase harvestable value of the raw materials and improve UCL’s recycling rate, while removing user input from the disposal of waste process.

The Design

The design emphasized user convenience by having a single inlet for all waste types. For this to be feasible a storage system was incorporated with a novel triangular slider mechanism used to separate items thrown in at the same time. Actuation systems were added to sort the items into their correct locations following use of material identification and contamination sensing specially-developed software. Stress and vibrational analysis of the design was performed to ensure that the design matched the current bins at UCL aesthetically and exceeded their current volumetric capacity safely and inexpensively. The results were very promising with the surface contamination software successfully quantifying contamination at high accuracy while material type classification was achieved with >96% accuracy.

Back to top