Precise point positioning for mobile robots

2013 IEEE/RSJ International Conference onIntelligent Robots and Systems (IROS)November 3-7, 2013. Tokyo, JapanPrecise Point Positioning for Mobile Robots using Software GNSSReceiver and QZSS LEX SignalTaro Suzuki, and Nobuaki Kubo, Member, IEEEAbstract— This paper describes outdoor localization for amobile robot using precise point positioning (PPP) based on theQuasi-Zenith Satellite System (QZSS) L-band Experiment(LEX) signal. For autonomous navigation applications, areal-time kinematic (RTK) global positioning system (GPS)technique is widely used to estimate user position withhigh-precision accuracy in real time. However, RTK-GPSrequires a reference station, and there are data acquisition costsinvolved in estimating the position.

Our approach correctsposition error by applying PPP using the QZSS LEX message.PPP can estimate a single receiver position without any referencestation or baseline, through use of satellite position fixing andclocks. We developed a method for extracting the QZSS LEXmessage in real time using a software GNSS receiver. We thenconstructed the PPP framework based on an LEX messagecontaining the satellite ephemeris and clock errors.

Finally, weconducted field experiments to evaluate the accuracy andprecision of our proposed method. The experimental resultsconfirmed that our method made a localization precision of 1.29m possible without using a GNSS reference station.I. INTRODUCTIONRecent developments in intelligent transportation system(ITS)–related technologies have actively promoted theincreased application of autonomous mobile systems inoutdoor environments.

Examples of such systems includeautomatic vehicle navigation systems and unmannedconstruction systems with autonomous constructionmachinery. Mobile mapping systems [1], for example, requirethe ability to carry out continuous and accurateself-positioning. In these applications, high-precisionsatellite-based positioning constitutes a fundamental part ofthe application [2]. Global navigation satellite systems(GNSSs), such as the U.S.-based Global Positioning System(GPS), have traditionally been used to estimate the position ofmobile robots in outdoor environments. The availability ofGNSSs is anticipated to improve as new positioning satellitesare launched.In applications that require high-precision positioning inreal time, a real-time kinematic (RTK) GPS technique isusually used [4].

RTK-GPS is one of the most precisepositioning technologies and can be used to obtaincentimeter-level accuracy of the position in real time byprocessing the carrier-phase measurements of GPS signals. Inthe Defense Advanced Research Projects AgencyGrand/UrbanChallenge involving driverless outdoor vehicle races, mostTaro Suzuki and Nobuaki Kubo are with the Tokyo University of MarineScience and Technology, 2-1-6, Etchujima, Koto-ku, Tokyo 135-8533, Japan.(phone and fax: +81-3-5245-7376; e-mail: tsuzuk0@kaiyodai.ac.jp).978-1-4673-6357-0/13/$31.00 ©2013 IEEEteams use RTK-GPS/inertial navigation system (INS)navigation.

In 2005, the Stanford University Racing Teamachieved autonomous running for over 200 km [3]. RTK-GPSis very popular; however, it requires GNSS base stations orsimultaneous observations by dual-frequency GPS receivers,which is inconvenient where cost is a consideration. Inaddition, the maximum distance of the user from the referencestation must be less than 20 km because of ionospheric error,which is dependent on the location [5].

Unlike single-referencestation RTK-GPS approaches, where the positioning accuracydecreases as the baseline increases, network RTK (NRTK)approaches ideally provide positioning with errorsindependent of the rover position within the network [5].However, NRTK also requires a communication link and costsare involved to receive the correction data in many cases.In recent years, precise point positioning (PPP) has beenanticipated as an alternative to RTK-GPS [6][9].

PPP canestimate a single receiver position without any referencestation or baseline through the use of satellite position fixesand clocks, which are previously determined. In comparisonwith common techniques, the costs can be reduced. However,PPP requires fixed satellite positions and clocks, and thequestion is how to get these correction data. In this paper, weuse the L-band Experiment (LEX) signal broadcasted by theQuasi-Zenith Satellite System (QZSS) to solve this problem.The first QZSS was launched from Japan in September 2010[12].

QZSS uses an asymmetric figure eight orbit, and hence,the elevation angle of QZSS is more than 75° forapproximately 8 h/d in Tokyo, Japan [13]. The LEX messageprovides GPS satellite clock and ephemeris errors forcorrecting the GPS positioning error via the QZSS satellite.This function allows the GPS accuracy to be improved.Unfortunately, the LEX messages cannot be received usingcurrent commercial GNSS receivers. In addition, for real-timeapplications, the LEX message needs be extractable forcomputing user positions in real time.

In this study, wedeveloped a framework for PPP using the QZSS LEXmessage for outdoor mobile robots. We also developed anovel technique for receiving the QZSS LEX message using asoftware GNSS receiver. In GNSS research, software-definedreceivers (SDR) are widely recognized and utilized for theirconfiguration flexibility and ease of use [10][11]. We wereable to construct the PPP system based on the QZSS LEXmessage using a software receiver and to confirm the positionestimation capability in an actual outdoor environment.II.

OVERVIEW OF PROPOSED METHODThe proposed method to estimate position using the QZSSLEX signal consists of (1) LEX message computation using a369software GNSS receiver, and (2) localization of mobile robotsusing PPP with LEX messages. Fig. 1 shows the system flowwith a PPP using the LEX messages. First, we developed thesoftware GNSS receiver to extract the LEX message. In thisstudy, rather than tracking the LEX signal, the LEX messagewas extracted with the aid of the conventional L1CA codephase and the Doppler signal obtained by the L1CA signaltracking. Using this method, each GPS satellite ephemeris errorand clock error could be easily used for PPP in a simpleconfiguration.

The precise satellite position and clock werecalculated by the PPP framework. We computed the location ofthe mobile robot and other unknown values using a Kalman filter.QZSSGPSGNSS Front EndLEX MessageComputationGPS ClockErrorEphemerisErrorSoftware receiverGNSS ReceiverSatelliteKalmanpositionFilterand clockcalculationPPP using LEXUser positionFigure 1. Overview of proposed positioning method.

There are two parts: (1)LEX message computation using a SDR, and (2) PPP with LEX messages.III. DECODING OF LEX MESSAGE USING SOFTWARE RECEIVERA. QZSS LEX SignalThe LEX signal, an augmentation signal in the 1278.75-MHzband broadcasted from QZSS, is anticipated to provide finepositioning up to the centimeter level [7][8]. Fig. 2 shows thestructure of the LEX message used for the broadcast of theaugmentation information. The LEX message consists of a49-bit header, 1695 bits of data, and 256 bits for errorcorrection using Reed-Solomon coding; the total size of onemessage is 2000 bits.

The actual data transfer volume for theaugmentation information is 1695 bps, with a basic structureof one message transmission per second. The augmentationinformation for the whole of Japan needs to be incorporatedinto 1695 bps. The LEX signal is overlaid by the ranging code,which has a period of 4 ms and a chip rate of 10,230 MChip/s.2000 bits / 1sHeader(49 bits)Data Part (1695 bits)Preamble (32 bits)PRN (8 bits)Message Type ID (8 bits)Alert Flag (1 bit)Figure 2. LEX message structure.Reed-SolomonCode (256 bits)The LEX message contains the signal health, ephemeris error,clock error, and ionospheric correction. The correction itemssuch as satellite ephemeris and clock errors are updated every12 s for high-speed corrections, and a low-speed correctionitem such as ionosphere error is updated every 30 min.B. Software GNSS ReceiverThe LEX signal utilizes code shift keying (CSK) [7].