No abstract

A novel sensor fusion approach between a wireless Ultra-Wideband (UWB) Indoor Positioning System (IPS) and an Inertial Navigation System (INS) is presented for real-time indoor navigation and tracking of Automated Guided Vehicles (AGVs) and Mobile Robots in factories and warehouses. AGV applications are particularly challenging since they require very low position errors and very high position update rates to be able to track the highly dynamic movements in real-time. The sensor fusion algorithm consists of two main blocks: the first is a delay compensation block that compensates the delayed results of position estimation coming from the IPS, while the second block is a multi-rate Extended Kalman Filter (EKF) that combines dynamic models of the movements, delay-compensated position measurements from the IPS and measurements coming from the INS. To validate the proposed sensor fusion approach, extensive tests have been performed on a real-time platform using a combination of an Ubisense UWB IPS and an INS based on two low-cost MEMS accelerometers and one gyroscope mounted on the Flanders Make Badminton Robot performing highly dynamic movements.The results show that the proposed sensor fusion approach is able to deliver a 1 kHz position update rate in realtime operation with a position error standard deviation of 3.7 cm for linear movements and 1.7 degrees for rotational movements. These are significant performance improvements compared to a stand-alone Ubisense UWB indoor positioning system or a standalone inertial navigation system.

This article presents sensor fusion techniques for ultra-wideband-based localization to achieve sufficient accuracy and robustness for the control of vehicles in an industrial environment. We propose two outlier detection methods in combination with an extended Kalman Filter, and present experimental validation where 10 cm accuracy is achieved even in difficult NLOS conditions.

No abstract

No abstract