Automatic monitoring of bins

Abstract: When a pile of scrap at Stena recycling has reached the maximum height, the loader operators need to drive around the facility to check whether or not the pile has to be emptied. This causes a lot of unnecessary driving, resolving in wear to the loaders and loss of time, but mostly the unnecessary use of diesel. In this paper different sensors is evaluated and tested against each other to find a system that can measure the volume of the pile. The sensors used is IFM O3D303 and the Kinect V1 sensor. These sensors are then used with two different volume calculation software, CloudCompare and 3DF Zephyr. The goal of the project is to get a system that can get a volume with an accuracy within 2.5 % of the actual volume. If a volume can be calculated from the scrap piles, then it is also possible to know when the pile should be cleared. Since the facility is a recycling plant, there is a lot of metal dust that is gathering in the indoor part of the facility. The sensors being used, should be able to work in this environment. To do that the sensor needs to have an IP certification higher than IP 55. The distance of measuring the pile can be more than 10 m, so the sensors should also be able to handle this requirement. From the sensors a point cloud and a mesh were captured, two data formats that contain information of where every point on the object is located in a three-dimensional space. Every system needs to be calibrated to suit the specific application. The results show that, even with a basic sensor such as the Kinect V1, an accurate volume of a scrap pile could, with calibration, be calculated. However, the cheap sensor does have limits, unlike the IFM which can cope with all the requirements. This report shows how the two sensors compare to each other and how the different software, combined with the sensors, can be used to calculate a volume of a pile of scrap.