Comparison on Cloud Image Classification for Thrash Collecting LEGO Mindstorms EV3 Robot

Abstract

The world today faces the biggest waste management crisis due to rapid economic growth, congestion, urban planning issues, devastating negative symptoms and political affairs. In addressing this waste management problem, many methods of solving waste management have proven not to be as planned. In this high technology era, the innovation of humanoid robots is found to be helpful to support the everyday human life. The industry is gearing towards automation to increase productivity at the same time will improved quality of life to local communities. Therefore, in this paper Thrash Collecting Robot (TCR) is proposed to help provide automatic control in thrash collection. The TCR, built on the LEGO Mindstorm EV3 robot, can distinguish between static and dynamic barriers, and can move according to the programming that has been created. TCRs are basically composed of sensors designed according to different requirements in order to detect dynamic barriers. TCR is one type of Cloud Robot that implements image processing techniques to identify the type of waste that has been collected. The concept of image processing built in TCR by using Cloud Representational State Transfer (REST API). This concept has been applied by Google Cloud API and Sighthound. This cloud services used machine vision techniques to identify and classify the type of thrash images; whether it is plastic, metal or paper. Experiment results show that SightHound gives accurate result compared to Google Cloud in classifying thrash types.