Description

There is an increasing interest for low cost compact systems for true North seeking that do not include the availability of magnetometers or GPS sensors. Devices for the orientation and attitude for under water systems ROV (Remotely Operated Vehicle), avionic systems, drilling systems, portable weapon designation (i.e. binocular) are examples in which compactness and independence from magnetic fields or GPS are required for obvious operational needs: GPS signals do not reach under the sea surface, inside metal pipes for drilling there is no GPS signal and the magnetic field is heavily undermined by existing ferromagnetic elements.

Furthermore for strategic systems you don't want to rely upon GPS signal or satellite in general as this could be blocked for intentional or natural causes. Typically inertial sensors have a precision and cost directly proportional to their dimensions and therefore typically compact systems correspond to low precision.

There are different kinds of gyrocompass that use inertial sensors at solid state or with rotating mass technology.

The gyrocompass at rotating mass is essentially a gyroscope, or a wheel that because of the rotations tends to maintain its axis always with the same orientation in respect to an inertial reference system. The wheel is kept continuously rotating by an electric motor. Since Earth rotates, an observer on the terrestrial surface observes that the gyroscope axis performs a rotation every 24 hours, always pointing toward the same direction in respect to the fixed stars.

The gyrocompass which performs 6 degrees of freedom (three accelerometers and three gyroscopes) are typically of the strap-down kind and can use both gyroscopic sensors in solid state (such as FOG - Fiber Optic Gyroscope or MEMS - Micro Electrical Mechanical Systems, therefore without some moving parts) and rotating mass sensors such as DTG (Dynamically Tuned Gyroscope).

The invention in question is related in particular to gyrocompass with sensors at solid state but that has moving parts. In particular we are referring to known techniques of true North research, such as those reported by Watson in the patent US5272922A or in the article "Automatic north sensor using a fiber-optic gyroscope", Tomohiro Tanaka in Applied Optics, vol. 33 n.l, 1 January 1994 in which a gyroscope at solid state is made to rotate at low speed around its non sensitive axis on an horizontal plane to measure the angular velocity of Earth obtaining a sine signal whose maximum, minimum and zeros represent the cardinal positions.

The main limits to the solutions so far proposed are the following:

1) The electromechanical gyrocompasses at rotating mass have the downside of being very cumbersome, noisy and they need annual programmed maintenance to change the suspension liquid and also maintenance for the replacement of the sensitive element every five years.

2) The strap-down systems at six degree of freedom have the downside to be very expensive systems as to have high precision systems you need to use very expensive sensors. They are quite compact but definitely not suitable for the mentioned applications in the first part of the description.

3) Systems with solid state sensors have the advantage of being very compact, virtually without any programmed maintenance but they work only in static conditions with obvious limitations in possible applications.

Referring to other known techniques of true North research, United States Patent Application 20090119937 from Watson William "METHOD AND SYSTEM FOR HEADING INDICATION WITH DRIFT COMPENSATION" the solution claimed in such application is useful only when the system is stationary, while the solution claimed does not compute the full attitude in dynamic conditions.

Moreover in WO 201 1/146990 from IMDEX the solution defined in such application can be considered only a structural (no algebraic method is cited in such application) extension of the solution proposed by Watson (US 20090119937), because it use a sequence of fixed positions to determine the actual attitude of the whole sensor body.

Therefore a strap-down system is hereby submitted, made of a static part and a continuous rotating part, able to calculate independently (without the help of GPS or magnetic fields) attitude and direction in respect to true North, using a configuration at five degrees of freedom (three accelerometers and two gyroscopes) with sensors arranged and set on the fixed and the rotating part of the system as follows: two mutually orthogonal accelerometers (104 and 105) with sensitive axis on the horizontal plane and a gyroscope (1 10) with sensitive axis in the vertical direction in the fixed part of the system, plus the third accelerometer (1 11) with a sensitive axis on the horizontal plane and the second gyroscope (108) with direction of the sensitive axis on the horizontal plan in the rotating part of the system.

With the strap-down system in static condition, the continuous rotating gyroscope (108) measures the sine projection of the Earth speed on its axis surveying, during its rotation, the maximums, minimums and zeros corresponding to North, South, East and West. Such gyroscope has an automatic system of rotation bias removal. The static gyroscope (110) together with the two accelerometers (104 and 105) measures the trim of the strap-down system and through an algorithm helps maintain the direction found by the rotating gyroscope when the conditions are not static. The accelerometer in the rotating part measures the fixed accelerometers bias thanks to the rotation of the sensitive axis on the horizontal plane.

The implemented equations in the calculation device have been optimized for the determination of the attitude, of the angular velocity and of the gyroscopes and fixed and rotating accelerometers bias. For this purpose the fixed reference system S _B integral with the strap-down system and the rotating reference system S _R integral with the rotating part of it have been defined. In these reference systems the static gyroscope (1 10) is parallel to the Z _B axis, the two fixed accelerometers (104 and 105) are placed parallel to the axis XB and Y _B , the rotating gyroscope (108) is parallel to the rotating axis X _R while the rotating accelerometer is parallel to the Y _R axis and far from the rotating axis for quantity p.

Supposing that the strap-down system is in a state of rectilinear uniform motion, that the local reference system speed is negligible (and its effects can be compensated according to standard tables), that the scale factors of the sensors bring to a negligible global effect on the measure, the equations which define the actual surveyed measure appear to be dependent on the trim, the angular velocity and the rotating part rotation position given at any one instant by the encoder and the bias of each individual sensor, according to the classic relations of kinematics of the rigid body in a reference system NED (North East Down):

(201) a _Bm [ _X y] = f(B _Ba [ _Xy ], Θ , λ)

(202) u¾i ^• Bm[z]— f(B _Bo [z], G¾N-BJ ON, λ)

(203) aRmfyj = f(BR _a { _y ], VR. _N , VR. _N , OR, ON, λ)

(204) ORj.Rmfx]— f(B _Ro[x] , CORN R, OR, ON, λ) where we have indicated with:

• indexes B, R, N— > the reference systems which are fixed in relation to, respectively, the strapdown system, the rotating system and the local reference system NED;

· v _A B - > speed of the reference system A in respect to reference system B, expressed in the reference system B;

• G _R → rotating angle of the rotating part in respect to the fixed part given by the encoder (106);

• ¾ -) trim of the strap-down system in respect to the local reference system NED, that can be represented through the euler angles or, even more efficiently, with quaternions or matrix of trim;

• λ— > actual latitude;

• G>AB c ^{→ rea} l angular velocity of the reference system A in respect to the reference system B expressed in the reference system C;

· GOAB Cm ^→ surveyed angular velocity of the reference system A in respect to reference system B expressed in the reference system C;

• asm— surveyed acceleration in the reference system S;

• indexes _[xyz] — > selection of the parts in the axis x, y and z;

• B _Ba [xy] ₅ BRa[y], B _Bo [ _z ] ₅ B _R O[ _X ]— > Respectively, accelerometers bias on the axis B _x and B _y , accelerometers R _y bias, gyroscope B _z bias and gyroscope R _x bias;

• gNs - > local gravity vector, calculated using the formulas and tables which express the gravity in relation with latitude.

Amongst the possible configurations of the surveying system it is possible to consider a degraded version which includes the removal of horizontal accelerometers and therefore the removal from the set of equations of the relation (201) and therefore all the reference to accelerometers bias done in the next formulas. A higher performance version of the system is instead possible through the addition of a further horizontal gyroscope oriented along Y _R axis of the rotating sdr R. In this case a similar equation to (204) should be added, written for Y _R axis and the bias of this gyroscope inserted in the list of state variables of the estimate filter described below.

The above mentioned equations are used for the implementation of an estimate filter which has the task of calculating the actual value of the state:

X = {Ββ ₃ [χγ], BRafy], BB ₀ [Z]> BR ₀ [ _X ], Θ , <¾N B} through the definition of the error vector on the state:

£χ ⁼ [EBbax? ¾bay ₅ ^£Bra y ^£ Bboz ₅ ¾rox> £θΝ? £ ⁽ DBNB] where we have indicated with:

^Bbax, EBbay, £ _Br ay, E _B boz, ε _ΒΓΟ χ—> estimate errors on the bias of five inertial sensors ε¾Ν— > vector of the estimate error on the trim in NED axis ε _Ω ΒΝΒ — > vector of the estimate error on the angular velocity of the strapdown

The state vector is updated at any one instant supposing constant bias through the kinematics equation of trim update:

(205) Θ _Ν = f(0 _N , C0BN B) operation which is more efficiently feasible through the representation of the actual trim in quaternions or trim matrix.

Equation (205) brings to the definition of the update equation of the filter state: (206) έθΝ - f(0N ₅ ε _ω ΒΝβ)

At every step of calculation the update of the estimate with the optimum standard method is done once matrix Po of starting covariance of the state vector X is defined, with Q as the covariance matrix of the implemented model and R the covariance matrix of the measures coming from five sensors, as follows:

(207) Φ = exp(F _k )

(211) P _k+ i = P' _k+ i - K H P' _k+ i

(212) £ _x = K z _k where we have indicated with:

• P' _k and P _k → the covariance matrix of the state, respectively, predicted and corrected, at time k

• F _k — state matrix, obtained at any one instant of calculation by the equation (206) of update of the filter state

· H— > input matrix, calculated by using equations linearization

(201) - (204) in respect to the state ε _χ

• Zk — the vector of the inputs at time k, calculated by using equations (201) - (204) and the actual estimated state X

• £χ→ the actual error estimate, used to correct the filter state · exp(-)— > matrix exponential operator. The calculation device performs first the initialization at a null value of the surveying system state X and at an adequate value of the matrix Po of covariance of the errors on the implemented filter state.

Then the calculation device performs an update of the state at each step of calculation, chosen based on the frequency of generation of measures by the inertial sensors, executing the operations in the following order:

• state propagation, using equation (205)

• calculation of the gain matrix K and update of the covariance matrix of the error using the equations (207) - (211)

· calculation of the vector z of the filter inputs using the linearization of equations (201) - (204) in respect to the state error vector ε _χ

• state correction, using the result of the operation (212).

Figure 1 shows a block diagram of the system (100) where it is indicated how the various sensors and inputs are elaborated by an estimation block realized through Kalman filter (1 12).

Figure 2 shows a perspective drawing of the gyrocompass system plus attitude reference system (100) with part of gyroscope MEMS sensor in rotation (108). A base (101) of support, a signals calculation and power source card (102), a control card (103) containing the motor (107 block (1 12) a rotating joint (109), a system to fasten the motor to the base (101) through the support (112), an encoder (106) whose through shaft is connected to the motor shaft, a gyroscopic sensor (108) which needs to rotate integral with the shaft (113) of the motor (107).

Figure 3 shows a drawing with lateral view of the proposed solution in which the accelerometers (105) and (104), the gyroscope (1 10) of not rotating part and the motor shaft (1 13) are also indicated.

Figure 4 shows a drawing with an additional lateral view of the proposed solution in which the accelerometer in the rotating part (11 1) is indicated. The advantages of the proposed solution in respect to the traditional models are the following:

1) Reduction of the encumbrance for the applications for which space is a problem (avionics, under water ROV systems, drones etc.).

2) Lower cost since the reduction of the bias thanks to rotations can allow the use of sensors not of "inertial grade" or "navigation grade" but technologies such as such as MEMS in example or "tactical grade" gyroscopes such as DTG (Dynamically Tuned Gyroscope) or FOG (Fiber optic Gyroscope).

If worth mentioning after this description the main differences with the mentioned Watson (US 20090119937). The system described differs substantially from the ones indicated in document US 20090119937 by the continuity of the motion and of the data acquisition of the process board (103), characteristics that is common to all the claims.

Moreover the proposed solution has a continuously rotating part and the sensors are continuously acquired to estimate the attitude independently of the actual and changing bias of the inertial sensors. This is achieved by adding the sensors in the static part that to allow, together with the sensors of the continuous rotating part, the acquisition and the tracking of the actual attitude when the system is dynamically changing its attitude. There's no reason to add the sensors in the static part when the system is stationary because the attitude can be determined even using only the sensors in the rotating part of the designed system allowing the detection and the continuous tracking of the actual attitude.

Moreover the proposed patent differs substantially from the US 20090119937 for the way the solution is computed. Solution in US 20090119937 proposes a sequence of acquisition of acceleration and angular rate in different positions of the rotating part and the computation of the actual attitude and heading using an algebraic solution of the attitude determination problem, given that the actual unknown data to determine are the attitude and the sensors biases. The solution in the proposed patent is based on the continuous acquisition of the accelerometer and gyros output, to feed a continuous estimation algorithm that is more robust respect to the presence of sensor noise, bias and bias variation. While US 20090119937 defines a solution based, as mentioned before, on a series of static acquisition of sensors rotated in a given set of different directions. All the solutions by the proposed patent are based on a continuous acquisition of sensors to obtain, both in stationary conditions and in dynamic conditions the computation of the actual attitude despite to the presence of sensor biases and bias variation, adding and removing sensors in the various claims with the purpose to increase the performances in terms of precision or acquisition time, or other useful information to acquire.