US 7,228,230 B2Grant

System for autonomous vehicle navigation with carrier phase DGPS and laser-scanner augmentation

Issue Date:2007-06-05

•20 Claims

•21 Drawing Sheets

Abstract

A horizontal navigation system aided by a carrier phase differential Global Positioning System (GPS) receiver and a Laser-Scanner (LS) for an Autonomous Ground Vehicle (AGV). The high accuracy vehicle navigation system is highly demanded for advanced AGVs. Although high positioning accuracy is achievable by a high performance RTK-GPS receiver, the performance should be considerably degraded in a high-blockage environment due to tall buildings and other obstacles. The present navigation system is to provide decimetre-level positioning accuracy in such a severe environment for precise GPS positioning. The horizontal navigation system is composed of a low cost Fiber Optic Gyro (FOG) and a precise odometer. The navigation errors are estimated using a tightly coupled Extended Kalman Filter (EKF). The measurements of the EKF are double differenced code and carrier phase from a dual frequency GPS receiver and relative positions derived from laser scanner measurements.

Metadata

Assignee

Mitsubishi Denki Kabushiki Kaisha

Inventor

Rui Hirokawa

Application Information

Application Number:US 10/986,205

Filing Date:2004-11-12

Priority Date:2004-11-12

Art Unit:3661

Classifications

IPC:

G01C21/00

Patent Drawings (21 sheets)

Description

Background of the Invention

[0001] Field of the Invention

[0002] The present invention relates generally to navigational systems using inertial and satellite-based navigation.

[0003] The present invention includes the use of various technologies referenced and described in the documents identified in the following LIST OF REFERENCES, which are cited throughout the specification by the corresponding reference number in brackets:

List of References

[0004] [1] Paul de Jonge and Christian Tiberius. The LAMBDA method for integer ambiguity estimation: implementation aspects. LGR-Series 12, Delft University of Technology, August 1996.

[0005] [2] Rui Hirokawa, Koichi Sato, and Kenji Nakakuki. Design and evaluation of a tightly coupled GPS/INS using low cost MEMS IMU. In GNSS Symposium 2003, Tokyo, Japan, November 2003.

[0006] [3] Tetsuro Imakiiere, Yuki Hatanaka, Yohta Kumaki, and Atsusi Yamagiwa. Geonet: Nationwide GPS array of Japan. GIS development, 8(3), March 2004.

[0007] [4] J. Takiguchi and J. Hallam. A study of autonomous mobile system in outdoor environment (part 3 local path planning for a nonholonommic mobile robot by chain form). In IEEE International Vehicle Electronics Conference, Changchun, China, September 1999.

[0008] [5] K. Ohno and T. Tsubouchi. Outdoor navigation of a mobile robot between buildings based on dgps and odometry data fusion. In IEEE International Conference on Robotics and Automation, volume 2, pages 1978–1984, 2003.

[0009] [6] Sachin Modi. Comparison of three obstacle avoidance methods for an autonomous guided vehicle. Master's thesis, University of Cincinnati, 2002.

[0010] [7] P. J. G. Teunissen. The least-squares ambiguity decorrelation adjustment: A method for fast GPS integer ambiguity estimation. Journal of Geodesy, 70(1), 1995.

[0011] [8] Robert M. Rogers. Applied Mathematics in Integrated Navigation Systems. AIAA Education Series. AIAA, second edition edition, October 2003.

[0012] [9] S. Sukkarieh and E. Nebot. A high integrity IMU/GPS navigation loop for autonomous land vehicle application. IEEE Trans. on Robotics and Automation, 15(3):572–578, June 1999.

[0013] The entire contents of each reference in the above LIST OF REFERENCES is incorporated herein by reference.

Discussion of the Background

[0014] Developing an outdoor navigation system that enables a vehicle to go through many obstacles is a major challenge for autonomous ground vehicles with GPS receivers [4, 6]. Navigation systems integrating low-cost inertial sensors and high performance GPS receivers will be applied in wide variety of mobile systems. The navigation system based on IMU or odometry coupled with GPS is widely used [9, 5]. However, performance is considerably degraded in a high-blockage environment due to tall buildings and other obstacles. A tightly coupled GPS/INS [2] has several advantages over a loosely coupled system such as better blunder detection of GPS pseudorange and higher positioning accuracy, especially under poor satellite visibility.

Summary of the Invention

[0015] According to an aspect of the present invention, there is provided a navigational device for determining the position and the heading of an object, comprising a navigation calculation device configured to calculate the position and the heading of the object based on an output of a velocity detecting device and an output of a yaw rate detecting device; and an estimator configured to estimate, based on a carrier phase and a pseudorange received from a global positioning satellite, (1) an error of the velocity detecting device, (2) an error of the yaw rate detecting device, (3) a position error and a heading error of the object, and (4) an integer-valued bias of a carrier, wherein the navigation calculation device is configured to update the position and the heading of the object based on the position error and the heading error estimated by the estimator.

[0016] In addition, an embodiment of the present invention is a terrestrial vehicle having embedded therein the navigational device described above.

[0017] Further, according to another aspect of the present invention, there is provided a navigational method of determining a position and heading of an object, comprising: calculating the position and the heading of the object based on an output of a velocity detecting device and an output of a yaw rate detecting device; estimating, based on a carrier phase and a pseudorange received from a global positioning satellite, (1) an error of the velocity detecting device, (2) an error of the yaw rate detecting device, (3) a position error and a heading error of the object, and (4) an integer-valued bias of a carrier; and updating the position and the heading of the object based on the estimated position error and the estimated heading error.

[0018] In another embodiment of the present invention, there is provided, a navigational system for determining a position and heading of a vehicle using inertial and satellite navigation, comprising: (1) a velocity detecting device configured to detect a velocity of the vehicle; (2) a yaw rate detecting device configured to detect a yaw rate of the vehicle; (3) a landmark database configured to store position data of a known landmark; (4) a range measuring device attached to the vehicle, the range measuring device configured to measure a range from the vehicle to the known landmark; (5) a navigation calculation device configured to calculate the position and the heading of the autonomous vehicle based on the velocity detected by the velocity detecting device and the yaw rate detected by the yaw rate detecting device; and (6) an estimator configured to estimate, based on a carrier phase and a pseudorange received from a global positioning satellite, an error of the velocity detecting device, an error of the yaw rate detecting device, a position error and a heading error of the object, and an integer-valued bias of a carrier, wherein the navigation calculation device is configured to update the position and the heading of the object based on the position error and the heading error estimated by the estimator.

[0019] In addition, another embodiment of the present invention is a terrestrial vehicle equipped with the navigation system of described above.

Brief Description of the Drawings

[0020] A more complete appreciation of the invention and many of the attendant advantages thereof will be readily obtained as the same becomes better understood by reference to the following detailed description when considered in connection with the accompanying drawings, wherein:

[0021] FIG. 1 illustrates a dynamical model of a terrestrial vehicle;

[0022] FIG. 2 illustrates a GPS/DR navigation system according to an embodiment of the present invention;

[0023] FIG. 3 illustrates a landmark update;

[0024] FIG. 4 illustrates a GPS simulation model;

[0025] FIG. 5 illustrates a horizontal navigation course used in a simulation;

[0026] FIG. 6 illustrates a satellite sky plot on Jul. 6, 2004;

[0027] FIG. 7 illustrates the velocity, heading, and yaw rate of a vehicle during simulation;

[0028] FIG. 8 illustrates the sensor errors estimated by an extended Kalman filter;

[0029] FIG. 9 illustrates the position error estimated by an extended Kalman filter;

[0030] FIG. 10 illustrates a horizontal navigation course used in a simulation with landmark update;

[0031] FIG. 11 illustrates the sensor errors estimated by an extended Kalman filter in the case of landmark update;

[0032] FIG. 12 illustrates the position error estimated by an extended Kalman filter with and without landmark update;

[0033] FIG. 13 illustrates an autonomous ground vehicle (AGV);

[0034] FIG. 14 illustrates the horizontal position estimated by a Kalman filter during the field test;

[0035] FIG. 15 illustrates the NED position during the field test;

[0036] FIG. 16 illustrates the number of satellites and the estimated validity of the integer ambiguity;

[0037] FIG. 17 illustrates the vehicle path and a plot of the measured range of the laser scanner during the field test;

[0038] FIG. 18 illustrates the measurement data for the landmark update during the field test;

[0039] FIG. 19 illustrates the calculated position update with and without the landmark update during the field test; and

[0040] FIG. 20 illustrates a method according to an embodiment of the present invention.

Description of the Peferred Embodiments

[0041] The navigation system consists of horizontal strapdown navigation calculation and the extended Kalman filter (EKF). FIG. 1 shows a horizontal dynamics model of the vehicle. The strapdown calculation in a local horizontal frame is performed by position and heading (yaw angle) updates using rate gyro, e.g., a low-cost Fiber Optic Gyro (FOG), and the odometer input compensated by the EKF. A Micro Electro Mechanical Systems (MEMS) gyro and a vibrating gyro may also be used. The position of the vehicle is defined as the phase center of a GPS antenna, and the position dynamics in local NED (North-East-Down) frame is defined as follows,

[0042]

\begin{matrix} [\begin{matrix} \dot{N} \\ \dot{E} \end{matrix}] = R (ψ) [\begin{matrix} v_{x} \\ v_{y} \end{matrix}]; \dot{D} = 0 & (1) \\ R (ψ) = [\begin{matrix} \cos ψ & - \sin ψ \\ \sin ψ & \cos ψ \end{matrix}] & (2) \end{matrix}

where v_xand v_yare the body-frame velocity defined as follows,

[0043]

\begin{matrix} [\begin{matrix} v_{x} \\ v_{y} \end{matrix}] = [\begin{matrix} V_{s} \\ b_{r} r_{s} \end{matrix}] . & (3) \end{matrix}

V_sand r_sare the compensated velocity measured by the odometer and the yaw rate measured by FOG, respectively. b_ris the horizontal distance between the rear wheel axis and the GPS antenna phase center.

[0044] The compensated velocity and yaw rate are calculated by,

[0045]

\begin{matrix} V_{s} = \frac{1}{1 + e_{od}} V_{od} & (4) \\ r_{s} = \frac{1}{1 + e_{g}} (r - b_{g}) & (5) \end{matrix}

where V_odis the odometer velocity, r is the yaw rate, e_odis the odometer scale factor error, e_gis the gyro scale factor error, and b_gis the gyro bias error. The heading dynamics are modeled by,

[0046]

\begin{matrix} \frac{ⅆ}{ⅆ t} ψ = r_{s} & (6) \end{matrix}

The strapdown calculation is performed based on the integration of equations (1)–(6). Note that the velocity and yaw rate used in equations (1)–(6) may be uncompensated or compensated by the sensor error signal estimated by the Kalman filter, as shown in FIG. 2.

[0047] The EKF estimates states x=[p s a] consisting of position errors p=[n e d]^T(north, east, down), sensor errors s (heading error ε, odometer scale factor error e_v, gyro scale factor e_g, gyro bias b_g) and a float ambiguity vector α of double differenced (DD) carrier phase. The order of states depends on the number of tracked satellites. For a 12 channel dual frequency receiver, it is no more than 29 states.

[0048] FIG. 2 shows a block diagram of a navigation system, including the EKF 209, according to an embodiment of the present invention. The FOG/odometer sensors 208 provide the yaw rate and the velocity to the strapdown navigation unit 210. A laser scanner 207 provides relative position information to the EKF 209. As described below, the landmark database 206 provides geometric data of at least one known landmark. The GPS receiver 201, the DD calculation unit 202, and the satellite orbit calculation unit 204 provide the satellite navigation information to the EKF 209. As discussed in more detail below, the LAMBDA ambiguity resolution unit 203 and the cycle-slip detection unit 205 resolve ambiguity in the integer-valued bias of the carrier.

[0049] The Kalman filter 209 is implemented in feedback form so that the error in strapdown navigation unit 210 and that of sensor inputs can be compensated. The dynamics of the states x are represented by the linear differential equation,

[0050]

\begin{matrix} \frac{ⅆ}{ⅆ t} x = Fx + w & (7) \end{matrix}

where F is the system matrix and w is the process noise vector.

[0051] The system matrix F is derived from the system error dynamics, which are represented by the equations shown below,

[0052]

\begin{matrix} [\begin{matrix} \dot{n} \\ \dot{e} \end{matrix}] = \frac{\partial R (ψ)}{\partial ψ} [\begin{matrix} v_{x} \\ v_{y} \end{matrix}] \in + R (ψ) [\begin{matrix} v_{x} e_{v} \\ v_{y} e_{g} \end{matrix}] + b_{r} [\begin{matrix} - \sin ψ \\ \cos ψ \end{matrix}] b_{g} + w & (8) \\ \dot{\in} = {re}_{g} + b_{g} + w & (9) \\ {\dot{b}}_{g} = \frac{- 1}{τ_{g}} b_{g} + w & (10) \end{matrix}

where τ_gis the correlation time of the gyro bias.

[0053] The error dynamics of other states including the altitude, float ambiguities, and sensor inputs are formulated as random-walk process models.
{dot over (X)}_i=W_i (11)
The dynamics of the Kalman filter are represented in discrete form. The transition matrix of the system Φ is calculated by,

[0054]

\begin{matrix} Φ_{k} = I + τ F + \frac{τ^{2}}{2} F^{2} & (12) \end{matrix}

where τ is the sampling time.

[0055] The Kalman filter calculation consists of the time propagation and the measurement update. The time propagation for covariance matrix P is defined as,
P_k+1⁻=Φ_kP_k⁺Φ_k^T+Q_d (13)
where Q_dis the process noise matrix.

[0056] The measurement update of the EKF is defines as,
K=P_k⁻H^T(HP_k⁻H^T+R)⁻¹ (14)
P_k⁺=(I−KH)P_k⁻ (15)
x_k=KΔz (16)
where H is measurement matrix, K is Kalman gain matrix, R is measurement noise matrix, and Δz is the measurement residual.

[0057] The position, heading angle, sensor errors, and the float ambiguities are updated by as follows,

[0058]

\begin{matrix} {[\begin{matrix} N \\ E \\ D \\ ψ \\ e_{od} \\ e_{g} \\ b_{g} \\ a_{11} \\ ⋮ \\ a_{1 m} \\ a_{21} \\ ⋮ \\ a_{2 m} \end{matrix}]}_{k} = {[\begin{matrix} N \\ E \\ D \\ ψ \\ e_{od} \\ e_{g} \\ b_{g} \\ a_{11} \\ ⋮ \\ a_{1 m} \\ a_{21} \\ ⋮ \\ a_{2 m} \end{matrix}]}_{k} + {[\begin{matrix} x_{1} \\ x_{2} \\ x_{3} \\ x_{4} \\ x_{5} \\ x_{6} \\ x_{7} \\ x_{8} \\ ⋮ \\ x_{7 + m} \\ x_{8 + m} \\ ⋮ \\ x_{7 + 2 m} \end{matrix}]}_{k} & (17) \end{matrix}

[0059] where m is the number of observed satellites minus one.

[0060] The two different measurements, the DD range by GPS, and the relative range by the laser-scanner are used. The measurement matrix H and the measurement residual Δz are defined as shown below. The dual-frequency DD carrier phases and code phases are used as measurements of the filter.

[0061] After an integer ambiguity is obtained successfully, the order of EKF is decreased to be seven, and only the ambiguity resolved dual-frequency double differenced (DD) carrier phase are used as measurement.

[0062] The measurement matrix H of EKF is defined as,

[0063]

\begin{matrix} H = [\begin{matrix} A C_{n}^{e} & 0_{m \times 4} & 0_{m \times m} & 0_{m \times m} \\ A C_{n}^{e} & 0_{m \times 4} & λ_{1} I_{m} & 0_{m \times m} \\ A C_{n}^{e} & 0_{m \times 4} & 0_{m \times m} & λ_{2} I_{m} \end{matrix}] & (18) \end{matrix}

where C_n^eis the direction cosine matrix between the local horizontal frame and the ECEF frame, A is the geometric projection matrix for each satellite, λ₁, λ₂are the wave length of GPS L1 and L2, respectively.

[0064] The measurement matrix R for the GPS measurement update is highly correlated because it is based on DD observations. The measurement residual Δz is defined as,

[0065]

\begin{matrix} Δ z = [\begin{matrix} Δ ρ_{c} - Δ ρ_{C 1} \\ Δ ρ_{c} - Δ ρ_{L 1} - λ_{1} a_{1} \\ Δ ρ_{c} - Δ ρ_{L 2} - λ_{2} a_{2} \end{matrix}] & (19) \end{matrix}

where Δρ_cis the DD geometric range, Δρ_c1, Δρ_L1and Δρ_L2are the measured DD code and carrier phase, and a₁, and a₂are the DD ambiguities for L1 and L2.

[0066] After the integer ambiguities are resolved successfully, the DD carrier phase measurements are used with the resolved integer ambiguities {hacek over (a)}.

[0067]

\begin{matrix} Δ z = [\begin{matrix} Δ ρ_{c} - Δ ρ_{L 1} - λ_{1} {\overset{⋓}{a}}_{1} \\ Δ ρ_{c} - Δ ρ_{L 2} - λ_{1} {\overset{⋓}{a}}_{2} \end{matrix}] & (20) \end{matrix}

[0068] The relative position measured by a laser scanner is also used as an additional measurement to compensate position and heading error. The corner landmark, which has a known position stored in the landmark position database, is used to estimate the error by measuring the relative position between the landmark and the laser scanner sensor.

[0069] The line landmark is composed by connecting the two successive corner landmarks. The relative range between the line landmark and the scanner can also be measured. The relative position cannot be measured for the line landmark, but the relative angle can be measured. The corner landmark update combining the line landmark is used for computational efficiency. FIG. 3 shows the definition of the landmark update.

[0070] The range r_abetween the corner landmark and the laser scanner is measured by the laser scanner. The position of the corner landmark can be detected by the discontinuity of the range. The direction in the body frame θ_ais also measured by the sensor. The range vector between the laser scanner and the corner landmark represented in the body frame and the navigation frame are defined as,

[0071]

\begin{matrix} {\overline{r}}_{a}^{b} = {\overline{r}}_{a} [\begin{matrix} \cos θ_{a} \\ \sin θ_{a} \end{matrix}] \equiv [\begin{matrix} r_{ax}^{b} \\ r_{ay}^{b} \end{matrix}] & (21) \\ {\overline{r}}_{a}^{n} = r_{a} [\begin{matrix} \cos (θ_{a} + ψ) \\ \sin (θ_{a} + ψ) \end{matrix}] \equiv [\begin{matrix} r_{ax}^{n} \\ r_{ay}^{n} \end{matrix}] & (22) \end{matrix}

[0072] The position of the sensor is calculated by the vehicle position and the heading.

[0073]

\begin{matrix} {\overline{p}}_{ls} = [\begin{matrix} N \\ E \end{matrix}] + b_{s} [\begin{matrix} \cos ψ \\ \sin ψ \end{matrix}] \equiv [\begin{matrix} N_{ls} \\ E_{ls} \end{matrix}] & (23) \end{matrix}

where b_sis the distance between the GPS antenna and the laser scanner.

[0074] The relative range r_cand the direction in the body frame θ_care calculated by,

[0075]

\begin{matrix} r_{c} = \sqrt{(N_{l m, k} - N_{ls}^{2}) + {(E_{l m, k} - E_{ls})}^{2}} & (24) \\ θ_{c} = \tan^{- 1} (\frac{N_{l m, k} - N_{ls}}{E_{l m, k} - E_{ls}}) - ψ & (25) \end{matrix}

where (N_lm,k; E_lm,k) is the horizontal position of the k-th corner landmark.

[0076] The other ranging measurement r_bindicated in FIG. 3 is used to define the relative angle of the line-landmark. The direction of the range vector r_bis defined as the direction of the corner landmark added by the given offset angle Δθ. The reference point B shown in FIG. 3 is defined as the crossover point of the range vector r_band the line landmark.

[0077] The range vector r_bin the body frame is defined as,

[0078]

\begin{matrix} {\overline{r}}_{b}^{b} = r_{b} [\begin{matrix} \cos (θ_{a} + Δ θ) \\ \sin (θ_{a} + Δ θ) \end{matrix}] \equiv [\begin{matrix} r_{bx}^{b} \\ r_{by}^{b} \end{matrix}] & (26) \end{matrix}

[0079] The heading angle of the line landmark ψ_lnis calculated by adding the corner landmark direction θ_Sand the known angle Δθ.

[0080]

\begin{matrix} {\hat{ψ}}_{l n} = \tan^{- 1} (\frac{r_{b} \sin (θ_{a} + Δ θ) - r_{a} \sin θ_{a}}{r_{b} \cos (θ_{a} + Δ θ) - r_{a} \cos θ_{a}}) + ψ & (27) \end{matrix}

Δθ is preferably defined as 30 deg.

[0081] The measurement residual Δ_zis defined as,

[0082]

\begin{matrix} Δ z = [\begin{matrix} {\hat{r}}_{a} - r_{c} \\ \hat{θ} - θ_{c} \\ {\hat{ψ}}_{l n} - ψ_{c} \end{matrix}] & (28) \end{matrix}

where ψ_cis the heading angle of the line landmark retrieved from the landmark database.

[0083] The measurement matrix H is defined as,

[0084]

\begin{matrix} H = [\begin{matrix} - \frac{r_{anx}}{r_{a}} & - \frac{r_{any}}{r_{a}} & 0 & - \frac{r_{aby} b_{s}}{r_{a}} & 0_{1 \times (3 + 2 m)} \\ \frac{r_{any}}{r_{a}^{2}} & - \frac{r_{anx}}{r_{a}^{2}} & 0 & - \frac{r_{abx} b_{s}}{r_{a}^{2}} - 1 & 0_{1 \times (3 + 2 m)} \\ 0 & 0 & 0 & - 1 & 0_{1 \times (3 + 2 m)} \end{matrix}] & (29) \end{matrix}

[0085] The integer ambiguities vector {hacek over (a)} is resolved by the LAMBDA method [7, 1]. In the LAMBDA method, the integer least square problem shown below is solved, and the two best candidates of the integer ambiguities sets can be obtained.

[0086]

\begin{matrix} \hat{a} = \arg \min_{a \in Z^{n}} {(\hat{a} - a)}^{T} Q_{\hat{a}}^{- 1} (\hat{a} - a) & (30) \end{matrix}

where Q_â is the covariance of the float ambiguities, which is estimated by the EKF described above. Q_â is highly correlated because of the double differencing of ambiguities. Therefore, a direct search is ineffective. In the LAMBDA method, the de-correlation called Z-Transformation is performed before the ambiguity search. The search process can be considerably improved by the de-correlation process.

[0087] The integer ambiguities having the least square norm are verified using the ratio test and the residual test. The threshold level of the ratio test is defined as a function of the number of measurements and the confidence level. The 99% confidence level of the F-distribution is preferred.

[0088] FIG. 20 illustrates a method of determining a position and heading of an object according to an embodiment of the present invention. In step 2001, the position and the heading of the object is calculated using the dynamical equations:

[0089]

\frac{ⅆ ϕ}{ⅆ t} = r; \frac{ⅆ N}{ⅆ t} = V \cos ϕ - br \sin ϕ; \frac{ⅆ E}{ⅆ t} = V \sin ϕ + br \cos ϕ; \frac{ⅆ D}{ⅆ t} = 0,

wherein N, E, and D are components of the position, φ is the heading, V is a velocity measured by the velocity detecting device, r is a yaw rate measured by the yaw rate detecting device, t is time, and b is a predetermined distance. Next, in step 2002, a float ambiguity vector of double-differenced carrier phase is estimated using the extended Kalman filter and ambiguity in the integer-valued bias of the carrier is resolved using a Lambda method, based on the estimated float ambiguity vector of double-differenced carrier phase. In step 3003, based on a carrier phase and a pseudorange received from the global positioning satellite, the following values are estimated: (1) an error of the velocity detecting device, (2) an error of the yaw rate detecting device, (3) a position error and a heading error of the object, and (4) an integer-valued bias of a carrier. In step 2004, a relative range between a known landmark and the object is estimated based on a range measured by a range measuring device provided on the object and stored position data of the known landmark. In addition, the observation error is estimated based on the estimated relative range. Finally, in step 2005, the position and the heading of the object is updated based on the estimated position error and the estimated heading error.
Simulation

[0090] A design evaluation of embodiments of the navigation system and method of the present invention was performed by numerical simulation. FIG. 4 shows the simulation model including the various error sources. A horizontal path shown in FIG. 5 going through some tall buildings is used in the simulation. The GPS orbits are simulated using the almanac of the GPS week 1278. The nominal satellites masking angle is selected as 10 degrees.

[0091] A partial GPS satellite outage simulating the blockage by the tall building is considered in the first case. The blockage is represented by the 45 deg mask angle at t=50 s. The sky-plot of the observable satellites is shown in FIG. 6. The number of observed satellites is 10, but only 3 satellites are observed after the blockage. In this case, the conventional RTK-GPS can not output the fixed position. The tightly coupled GPS/DR navigation system here can perform the measurement update by the DD phase and code even if the number of satellite is less than four.

[0092] The sensor errors for the simulation are given as,
ε=0.2 deg, e_v=5%; e_g=5%; b_g=0.1 deg/s; (31)
FIG. 7 shows the velocity, heading, and yaw rate of the vehicle for the simulation. The sensor errors estimated by the EKF are shown in FIG. 8. The estimated errors converged to the true value. The error of the position by the EKF is shown in FIG. 9. For comparison, the error of position by the conventional DR navigation system loosely coupled with RTK-GPS is also shown in the same figure. The position error by the tightly coupled system is much smaller than the loosely coupled system. In second case, a full satellite blockage is simulated at t=50 s. In this case, the landmark update is effectively used. The laser scanner is assumed to have the maximum range of 30 m, range accuracy of 4 cm, and angle resolution of 1 deg. The landmark update is performed ahead of the corner landmark if the range is less than 30 m as shown in FIG. 10. FIG. 11 shows the estimated sensor errors by the EKF with landmark measurement update indicated as solid line and the true sensor errors indicated as dash line. The estimated errors converged to the true value. FIG. 12 shows the positioning errors with and without landmark updates. The position error is small even if no GPS measurements are available.
Field Test

[0093] An autonomous ground vehicle shown in FIG. 13 was employed for field tests. This AGV was developed as a surveillance vehicle that autonomously patrols in a certain area. It is equipped with a laser scanner, SICK LMS291 having 35 mm range accuracy and 1 deg angle resolution; Omni-Directional Video (ODV) camera for surveillance; a dual frequency GPS receiver, Ashtech Z-Xtreme; a FOG having accuracy 1 deg/h; and a high resolution odometer. The real-time guidance and GPS/DR navigation of the vehicle is performed on the onboard RT-Linux computer. The data analysed herein was collected in Kamakura, Japan on Jul. 6, 2004. The dual frequency GPS data observed in the base station of the GEONET [3] at Fujisawa is used for the differential correction. The baseline length is about 5 km. The observed satellites are almost the same as shown in FIG. 6. The ambiguity-fixed position by the RTK-GPS receiver was available only 40% of the time, and 32% of the time had no solution.

[0094] The current onboard real-time implementation is based on the loosely coupled navigation. The tightly coupled navigation solutions were obtained by a post processing calculation. FIG. 14 and FIG. 15 show the calculated position result of the field test. FIG. 16 shows the number of observed satellites and the estimated validity of the integer ambiguity. If the validity is one, the ambiguity is estimated to be fixed. For the tightly coupled navigation system, more than 70% of the position is based on the fixed solution. The centimetre-level accuracy was confirmed by comparing the calculated position by EKF with the fixed solution of RTK-GPS receiver. The position accuracy was confirmed by the residual analysis of DD carrier phase where the fixed solution of RTK-GPS is not available.

[0095] The landmark update is also performed at a second corner of the course. FIG. 17 shows the vehicle path and the plot of laser-scanner range at t=254 s and t=264.3 s. The plot of laser-scanner range is rotated by the heading of the vehicle to be matched with the path direction. This figure shows that the corner can be detectable with the range measurement by the laser-scanner.

[0096] FIG. 18 shows the examples of measured data by the laser-scanner at the second corner including the measured relative range and angle for the corner landmark, the measured slope for the line landmark.

[0097] The landmark update is performed without GPS measurement update because GPS/DR position shown in FIG. 14 already has high position accuracy. FIG. 19 shows the calculated position result with the landmark update. For comparison, the result without landmark update and GPS/DR position is also shown in the same figure. The calculated position result is nearly same as the GPS/DR position. With the corner and line landmark update, the position accuracy is improved when no GPS measurements are available.

[0098] A horizontal navigation system aided by a carrier phase DGPS and a Laser-Scanner for an AGV is designed and the performance is confirmed by numerical simulations and field tests. An embodiment of the present invention shows that decimetre-level positioning accuracy is achievable under poor satellite visibility by the proposed tightly coupled navigation system using carrier phase DGPS and LS augmentation.

Claims

What is claimed is:

1. A navigational device for determining a position and heading of an object, comprising:

a navigation calculation device configured to calculate the position and the heading of the object based on an output of a velocity detecting device and an output of a yaw rate detecting device; and

an estimator configured to estimate, based on a carrier phase and a pseudorange received from a global positioning satellite, (1) an error of the velocity detecting device, (2) an error of the yaw rate detecting device, (3) a position error and a heading error of the object, and (4) an integer-valued bias of a carrier,

wherein the navigation calculation device is configured to update the position and the heading of the object based on the position error and the heading error estimated by the estimator.

2. The navigational device of claim 1, wherein the estimator is configured (1) to estimate a relative range between a known landmark and the object based on a range measured by a range measuring device provided on the object and stored position data of the known landmark, and (2) to calculate an observation error based on the estimated relative range.

3. The navigational device of claim 2, wherein the estimator is configured to estimate the relative range based on geometrical data of at least two known landmarks and a known yaw angle of the range measuring device.

4. The navigational device of claim 1, wherein the navigation calculation device is configured to calculate the position and the heading of the object using the dynamical equations:

\frac{ⅆ ϕ}{ⅆ t} = r; \frac{ⅆ N}{ⅆ t} = V \cos ϕ - br \sin ϕ; \frac{ⅆ E}{ⅆ t} = V \sin ϕ + br \cos ϕ; \frac{ⅆ D}{ⅆ t} = 0,

5. The navigational device of claim 1, wherein the estimator comprises:

an extended Kalman filter configured to estimate a float ambiguity vector of double-differenced carrier phase; and

an ambiguity resolution device configured to resolve ambiguity in the integer-valued bias of the carrier, using a Lambda method, based on the float ambiguity vector of double-differenced carrier phase estimated by the extended Kalman filter.

6. The navigational device of claim 1, wherein the navigation calculation device is configured to calculate the position and the heading of the object based on an output of an odometer and an output of a rate gyro.

7. A terrestrial vehicle having embedded therein the navigational device of claim 1.

8. A navigational method of determining a position and heading of an object, comprising:

calculating the position and the heading of the object based on an output of a velocity detecting device and an output of a yaw rate detecting device;

estimating, based on a carrier phase and a pseudorange received from a global positioning satellite, (1) an error of the velocity detecting device, (2) an error of the yaw rate detecting device, (3) a position error and a heading error of the object, and (4) an integer-valued bias of a carrier; and

updating the position and the heading of the object based on the estimated position error and the estimated heading error.

9. The method of claim 8, wherein the estimating step comprises:

estimating a relative range between a known landmark and the object based on a range measured by a range measuring device provided on the object and stored position data of the known landmark; and

calculating an observation error based on the estimated relative range.

10. The method of claim 9, wherein the estimating step comprises:

estimating the relative range based on geometrical data of at least two known landmarks and a known yaw angle of the range measuring device.

11. The method of claim 9, wherein the estimating step comprises:

estimating the relative range based on a range measured by a laser scanner.

12. The method of claim 8, wherein the calculating step comprises:

calculating the position and the heading of the object using the dynamical equations:

\frac{ⅆ ϕ}{ⅆ t} = r; \frac{ⅆ N}{ⅆ t} = V \cos ϕ - br \sin ϕ; \frac{ⅆ E}{ⅆ t} = V \sin ϕ + br \cos ϕ; \frac{ⅆ D}{ⅆ t} = 0,

13. The method of claim 8, wherein the estimating step comprises:

estimating a float ambiguity vector of double-differenced carrier phase using an extended Kalman filter; and

resolving ambiguity in the integer-valued bias of the carrier, using a Lambda method, based on the estimated float ambiguity vector of double-differenced carrier phase estimated by the extended Kalman filter.

14. The method of claim 8, wherein the calculating step comprises:

calculating the position and the heading of the object based on outputs of an odometer and a rate gyro.

15. A navigational system for determining a position and heading of a vehicle using inertial and satellite navigation, comprising:

a velocity detecting device configured to detect a velocity of the vehicle;

a yaw rate detecting device configured to detect a yaw rate of the vehicle;

a landmark database configured to store position data of a known landmark;

a range measuring device attached to the vehicle, the range measuring device configured to measure a range from the vehicle to the known landmark;

a navigation calculation device configured to calculate the position and the heading of the vehicle based on the velocity detected by the velocity detecting device and the yaw rate detected by the yaw rate detecting device; and

an estimator configured to estimate, based on a carrier phase and a pseudorange received from a global positioning satellite, (1) an error of the velocity detecting device, (2) an error of the yaw rate detecting device, (3) a position error and a heading error of the vehicle, and (4) an integer-valued bias of a carrier,

wherein the navigation calculation device is configured to update the position and the heading of the vehicle based on the position error and the heading error estimated by the estimator.

16. The navigational system of claim 15, wherein the estimator is configured (1) to estimate a relative range between the known landmark and the vehicle based on a range measured by a range measuring device provided on the vehicle and position data of the known landmark stored in the landmark database, and (2) to calculate an observation error based on the estimated relative range.

17. The navigational system of claim 16, wherein the estimator is configured to estimate the relative range based on geometrical data of at least two known landmarks stored in the landmark database and a known yaw angle of the range measuring device.

18. The navigational system of claim 15, wherein the navigation calculation device is configured to calculate the position and the heading of the vehicle using the dynamical equations:

\frac{ⅆ ϕ}{ⅆ t} = r; \frac{ⅆ N}{ⅆ t} = V \cos ϕ - br \sin ϕ; \frac{ⅆ E}{ⅆ t} = V \sin ϕ + br \cos ϕ; \frac{ⅆ D}{ⅆ t} = 0,

19. The navigational system of claim 15, wherein the estimator comprises:

an extended Kalman filter configured to estimate a float ambiguity vector of double-differenced carrier phase; and

20. A vehicle comprising:

a propulsion system configured to propel the vehicle; and

a navigational system for determining a position and heading of the vehicle using inertial and satellite navigation, the navigational system including: