Benchmarking low-cost inertial measurement units for indoor localisation and navigation of AGVs

Abstract

In the last decade, the landscape of distribution and warehousing has drastically changed and is increasingly dominated by automated guided vehicles (AGVs) transporting palletised goods. The benefits that AGVs brought to the factory floor among which the decreasing labour costs, increasing safety, accuracy and productivity were also noticed by other sectors such as manufacturing, retail, e-commerce and, even military and healthcare. Nevertheless, the high initial investment and maintenance costs of these systems still remain bottlenecks for small and medium-sized enterprises (SMEs), slowing down their transition towards Industry 4.0. This demands for cost-effective AGVs equipped with low-cost exteroceptive and proprioceptive sensors for indoor navigation and localisation in highly dynamic environments. Therefore, this work presents the benchmarking of three low-cost and one medium-cost inertial measurement units (IMUs) characterised by an oft-recurring static analysis extended with a dynamic analysis. The former is based on the Overlapping Allan Variance (OAVAR) method and compares various sensor metrics such as white noise, bias offset and bias instability. The latter compares step and frequency response parameters of the IMUs, while being subjected to frequency ranges similar to those observed during AGV operation. In addition, these low and medium-cost IMUs are mounted on a mobile platform and applied in an indoor use case with a view to implement simultaneous localisation and mapping (SLAM) for AGVs in production sites. The influence of these IMUs on SLAM is investigated based on absolute and relative map metrics. Finally, it can be stated that with regard to AGV operation, the low-cost IMUs perform equally well compared to the medium-cost IMU.