Sensor systems for high speed intelligent sorting of waste paper in recycling

Sorting of paper into compatible grades is a necessary step prior to recycling. Current method of manually sorting is tedious, slow, and expensive. Hence, significant part of the waste paper stream is sent to the landfill. High speed automation of the sorting process will improve the cost efficiency and increase the amount of paper recycled significantly. There have been recent developments in automation of this process. Mechanisms for the distribution of the papers from a bale onto a moving conveyor and the pneumatic actuation to deflect papers into different streams are well established. Correct identification of the sample in real-time before the sample reaches the actuation station still remains a challenge since different types of paper and board samples are mixed in a waste stream, and has a lot of variation in terms of color, chemical composition, coating, and prints in black and white and color to different degrees. This is primarily due to the lack of satisfactory sensors and sensor fusion algorithms for sample identification in real-time. In this paper, we identify key parameters that must be measured, sensor design, and integration of the output from sensors to interpret the type of sample using a fuzzy inference system. Results show that the sensor system proposed is capable of identifying the samples at 90% accuracy. The sensor system can be integrated onto a conveyor and actuation system for automated sorting.