Municipal solid waste (MSW), generated at an unprecedented rate due to rapid urbanization and industrialization contains useful recyclable materials like metals, plastic, wood, etc. Recycling of useful materials from MSW in the developing countries is severely constrained by limited door-to-door collection and poor means of segregation. Recovery of recyclables is usually performed by waste pickers, which is highly risky and hazardous for their health. This paper reports the development of a robotic mobile manipulation system for automated sorting of useful recyclables from MSW. The developed robot is equipped with a thermal imaging camera, proximity sensor and a 5-DOF robotic arm. This paper presents an approach for sorting based on automated identification from thermographic images. The developed algorithm extracts keypoint features from the thermographic image and feeds into clustering model to map them into a bag-of-word vectors. Finally, Support Vector Machine (SVM) classifier is used for identifying the recyclable material. We used the developed algorithm to detect three categories of recyclables namely, aluminum can, plastic bottle and tetra pack from given thermographic images. We obtained classification rate of 94.3% in the tests. In future, we plan to extend the developed approach for classifying a wider range of recyclable objects as well as to incorporate motion planning algorithms to handle cluttered environments.

This content is only available via PDF.
You do not currently have access to this content.