Vision/UWB/IMU sensor fusion based localization using an extended Kalman filter

Most positioning technologies require some information about the immediate environment (e.g., whether the sensor is indoors or outdoors). Typical global positioning system (GPS) sensors cannot function inside buildings, so ultra-wideband (UWB) sensors may be used to determine location. However, beacons must be installed in at least three predetermined positions if UWB devices are to be used indoors. UWB sensors may also be used outdoors, but their operating range is limited as accuracy decreases with distance from the beacons. We propose a method for the fusion of UWB and other sensor data to calculate location with a low error rate both indoors and outdoors. UWB beacons were installed 1-m-apart to aid detection. The allowable separation between the tags and beacons was increased through use of a vision sensor, and colored balls were attached to the beacons to ease identification. The colored balls were detected using the circle Hough Transform (CHT) algorithm and a UWB tag. The CHT algorithm performs well when used to detect circles, but is computationally expensive. Therefore, we present a method that reduced this load by assigning regions of interest when UWB or inertial measurement unit (IMU) sensors were used. The tag position was calculated from the coordinates of the UWB beacons captured in an image and other positional data measured with the UWB sensor. Our proposed method, which includes the application of an extended Kalman filter (EKF), successfully calculated position with a greater accuracy than UWB alone. We intend to extend this concept and develop a method that will allow measurement of the tag position when the UWB beacons are moving.