Cooperative Simultaneous Localization and Mapping for Pedestrians using Low-Cost Ultra-Wideband System and Gyroscope
Christian Gentner, Markus Ulmschneider and Thomas Jost, German Aerospace Center (DLR), Germany
The proliferation of smartphones has made positioning technologies available to a wide range of users. For outdoor localization, global navigation satellite systems (GNSSs) are the most well-known and mostly used technologies for positioning. In open sky conditions, GNSSs provide sufficient position accuracy for most mass market applications. However, inside buildings or in urban canyons the GNSS positioning accuracy might be drastically reduced. In these situations, the received GNSS signals might be blocked, affected by multipath effects or received with low power. To enhance the positioning performance, different methods and sensor systems can provide position information to support or replace GNSSs. Most of the indoor positioning systems use local infrastructure like radio frequency identification (RFID) technologies, mobile communication base-stations, wireless local area network (WLAN) or ultra-wideband (UWB) systems. Using WLAN for indoor positioning is a common approach because WLAN infrastructure is widely deployed. On the other hand, UWB is a promising positioning system that has undergone massive research development in recent years. UWB systems use a large bandwidth which reduces the effect of multipath interference and facilitates the determination of times of arrival. Hence, UWB systems are a desirable solution for indoor positioning. In order to use UWB systems for positioning, UWB transmitters have to be placed on known locations. Without knowing the UWB transmitter positions, an accurate position estimate of the receiver is difficult or even impossible.
This paper presents a novel cooperative pedestrian localization algorithm using an UWB system without the necessity to have the prior information on the UWB transmitter positions. The proposed algorithm allows building up UWB transmitters at arbitrary positions. The novelty of the algorithm is to estimate the position of the UWB receiver and the UWB transmitters simultaneously, which can be interpreted as simultaneous localization and mapping (SLAM) with radio signals. To resolve ambiguities, we fuse the distance estimates of the UWB systems with heading information obtained from an IMU. Theoretically, the measurements of the IMU can be directly used in an inertial navigation system. However, the position calculation involves double integrations, and hence, even small measurement errors quickly cause a drift in the position solution. To avoid that, we only use heading measurements from the IMU which solely requires an alignment of the coordinate systems. As we are dealing with a relative positioning system, the derived algorithm requires prior information on the initial receiver position and moving direction only to define a local coordinate system. The proposed algorithm assumes no prior information on the UWB transmitter position. Hence, if the distance between an UWB transmitter and an UWB receiver is measured and estimated the first time, the corresponding UWB transmitter position has to be estimated solely using the distance estimate. The possible UWB transmitter position lies in a circular around the UWB receiver position with the radius of the estimated distance. As soon as the UWB receiver is moving, the estimations of the UWB transmitter positions slowly converge. To obtain accurate estimates of the UWB transmitter positions, the estimations of the UWB transmitter positions are shared between different UWB receivers. The concept of multiple UWB receivers allows to obtain an accurate estimate of the UWB transmitter position. An increased accuracy of the estimated UWB transmitter position relates directly to a more accurate position estimate of the UWB receiver position.
To verify the refined algorithm, we perform evaluations based on measurements. We use the DecaWave’s DW1000 UWB transceiver which enables cost effective real-time positioning with high accuracy in the order of 10cm in indoor and outdoor scenarios. In order to estimate the distance between the transmitters (anchors) and receiver(s) (tag(s)), we use a two way ranging method. The protocol contains four messages, two are sent by the tag, and two by the anchor. Afterwards, the distance is estimated based on the transmitting and receiving time stamps of tags and anchors. We evaluate the proposed algorithm based on measurements with moving pedestrians and fixed anchors with unknown positions. The measurements are carried out inside and outside of an office building. We placed UWB anchors on different positions located inside the building. The moving pedestrians carry a system consisting of the UWB tag, IMU and GPS receiver. The proposed algorithm estimates the position of the tag and anchors relatively, hence, works in a local coordinate system. For the measurements we assume that at least one pedestrian is starting outdoors before entering the building. Hence, the GPS receiver is used to initialize the position of the pedestrian outdoors in order to fix the coordinate system. The evaluations show that the horizontal positioning error of the developed positioning algorithm is around 2m for the first pedestrian. After multiple pedestrian shared there obtained map, the position accuracy of the pedestrian decreases below one meter. The evaluations show that an accurate position estimation of both the pedestrian and the anchors is possible without any prior knowledge on the anchor positions.
To conclude, using the developed SLAM algorithm allows building up an UWB indoor positioning system without measuring the locations of the UWB anchors. The UWB anchor positions are dynamically estimated using a cooperative SLAM approach.