Maassarani, Bilal2017-01-202018-01-212017-01-202017-01-18http://hdl.handle.net/10012/11227Weighing systems exist in various sizes and forms to meet the persistent demand for measuring the mass of objects. Current solutions do not offer a system that can dynamically weigh packages moving in a non-singulated and non-spaced fashion, common in automated settings. In such environments, currently items are singulated which results in a slower flow, increased cost and space requirements. In this work, we propose a design in which small-sized conveyors are mounted on load cells in a grid formation to dynamically weigh non-singulated objects that meet some minimal spacing requirements between them. In the design, moving packages are tracked with a vision system, and an algorithm is formulated to estimate mass based on filtered load cell outputs. Each element of the grid is modelled as a mass-spring-damper system in order to simulate the expected load cell output for the moving objects. A discrete time-variant low-pass filter is adopted from literature to filter the signal and an algorithm is devised to produce a mass estimate. A parameter estimation technique and a simple averaging method which ignore transients are implemented as well for performance comparison. The results are verified experimentally in two proof-of-concept experiments for a full scale prototype. When tuned properly, the time-variant filter succeeds in giving an estimate within a mean of 0.02% error of the rated load cell capacity at speeds up to 0.6 m/s. This is good performance since it does not exceed the rated error for the load cells. The other two estimation methods fail to meet the accuracy requirement at speeds above 0.4 m/s. Potential design considerations and concerns are discussed. Further development and testing is required before the machine can become legal-for-trade.enWeighingParameter EstimationFilteringTime-variant filteringWeighing machine designDynamic WeighingLowpass FilteringLoadcellMaster Thesis