Abstract
The key to machine automation, which is crucial to increasing production efficiency, is the use of sensorsfor product traceability. However, implementing product traceability significantly increases the cost of
an automated machine, driving the need for low-cost product traceability. Therefore, this research explores whether reliable position estimation could be achieved using only a single low-cost accelerometer sensor, where the exploration is performed on an automated line sorter.
The approach is to double integrate the measured acceleration in the direction of motion, causing the
position to drift over time. Therefore, the position is calibrated at certain positions using a Kalman filter
with intermittent observations and constant assumed bias. The calibration positions are obtained from
pattern-based position detection in the measured acceleration using signal processing.
The position estimation was evaluated for accuracy on experimental data by calculating the position error between the estimated position and the position measured by an additional ground truth data sensor used during the experiments, at three different velocities. It was concluded that the position estimation is accurate enough considering the total evaluation time. In addition, the percentage of successful position estimation evaluations was calculated, indicating that consistent position estimation can be achieved with only a single accelerometer. The high percentage of successful position estimation evaluations occurred at all the different sorter velocities, with a nearly linear position error as a function of these velocities. Based on these conclusions, it can be concluded that it is possible to perform reliable position estimation on an automated sorter using only a single accelerometer.
| Date of Award | 8 Jul 2024 |
|---|---|
| Original language | English |
| Supervisor | Alessandro Saccon (Supervisor 1) & Johan P.N. Freens (External coach) |