9-axis sensor fusion with Kalman filter

Add support for 9-axis gravity and linear-acceleration sensors
virtual orientation sensor using 9-axis fusion

Change-Id: I6717539373fce781c10e97b6fa59f68a831a592f
diff --git a/services/sensorservice/Fusion.cpp b/services/sensorservice/Fusion.cpp
new file mode 100644
index 0000000..56ac9f9
--- /dev/null
+++ b/services/sensorservice/Fusion.cpp
@@ -0,0 +1,431 @@
+/*
+ * Copyright (C) 2011 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <stdio.h>
+
+#include <utils/Log.h>
+
+#include "Fusion.h"
+
+namespace android {
+
+// -----------------------------------------------------------------------
+
+template <typename TYPE>
+static inline TYPE sqr(TYPE x) {
+    return x*x;
+}
+
+template <typename T>
+static inline T clamp(T v) {
+    return v < 0 ? 0 : v;
+}
+
+template <typename TYPE, size_t C, size_t R>
+static mat<TYPE, R, R> scaleCovariance(
+        const mat<TYPE, C, R>& A,
+        const mat<TYPE, C, C>& P) {
+    // A*P*transpose(A);
+    mat<TYPE, R, R> APAt;
+    for (size_t r=0 ; r<R ; r++) {
+        for (size_t j=r ; j<R ; j++) {
+            double apat(0);
+            for (size_t c=0 ; c<C ; c++) {
+                double v(A[c][r]*P[c][c]*0.5);
+                for (size_t k=c+1 ; k<C ; k++)
+                    v += A[k][r] * P[c][k];
+                apat += 2 * v * A[c][j];
+            }
+            APAt[j][r] = apat;
+            APAt[r][j] = apat;
+        }
+    }
+    return APAt;
+}
+
+template <typename TYPE, typename OTHER_TYPE>
+static mat<TYPE, 3, 3> crossMatrix(const vec<TYPE, 3>& p, OTHER_TYPE diag) {
+    mat<TYPE, 3, 3> r;
+    r[0][0] = diag;
+    r[1][1] = diag;
+    r[2][2] = diag;
+    r[0][1] = p.z;
+    r[1][0] =-p.z;
+    r[0][2] =-p.y;
+    r[2][0] = p.y;
+    r[1][2] = p.x;
+    r[2][1] =-p.x;
+    return r;
+}
+
+template <typename TYPE>
+static mat<TYPE, 3, 3> MRPsToMatrix(const vec<TYPE, 3>& p) {
+    mat<TYPE, 3, 3> res(1);
+    const mat<TYPE, 3, 3> px(crossMatrix(p, 0));
+    const TYPE ptp(dot_product(p,p));
+    const TYPE t = 4/sqr(1+ptp);
+    res -= t * (1-ptp) * px;
+    res += t * 2 * sqr(px);
+    return res;
+}
+
+template <typename TYPE>
+vec<TYPE, 3> matrixToMRPs(const mat<TYPE, 3, 3>& R) {
+    // matrix to MRPs
+    vec<TYPE, 3> q;
+    const float Hx = R[0].x;
+    const float My = R[1].y;
+    const float Az = R[2].z;
+    const float w = 1 / (1 + sqrtf( clamp( Hx + My + Az + 1) * 0.25f ));
+    q.x = sqrtf( clamp( Hx - My - Az + 1) * 0.25f ) * w;
+    q.y = sqrtf( clamp(-Hx + My - Az + 1) * 0.25f ) * w;
+    q.z = sqrtf( clamp(-Hx - My + Az + 1) * 0.25f ) * w;
+    q.x = copysignf(q.x, R[2].y - R[1].z);
+    q.y = copysignf(q.y, R[0].z - R[2].x);
+    q.z = copysignf(q.z, R[1].x - R[0].y);
+    return q;
+}
+
+template<typename TYPE, size_t SIZE>
+class Covariance {
+    mat<TYPE, SIZE, SIZE> mSumXX;
+    vec<TYPE, SIZE> mSumX;
+    size_t mN;
+public:
+    Covariance() : mSumXX(0.0f), mSumX(0.0f), mN(0) { }
+    void update(const vec<TYPE, SIZE>& x) {
+        mSumXX += x*transpose(x);
+        mSumX  += x;
+        mN++;
+    }
+    mat<TYPE, SIZE, SIZE> operator()() const {
+        const float N = 1.0f / mN;
+        return mSumXX*N - (mSumX*transpose(mSumX))*(N*N);
+    }
+    void reset() {
+        mN = 0;
+        mSumXX = 0;
+        mSumX = 0;
+    }
+    size_t getCount() const {
+        return mN;
+    }
+};
+
+// -----------------------------------------------------------------------
+
+Fusion::Fusion() {
+    // process noise covariance matrix
+    const float w1 = gyroSTDEV;
+    const float w2 = biasSTDEV;
+    Q[0] = w1*w1;
+    Q[1] = w2*w2;
+
+    Ba.x = 0;
+    Ba.y = 0;
+    Ba.z = 1;
+
+    Bm.x = 0;
+    Bm.y = 1;
+    Bm.z = 0;
+
+    init();
+}
+
+void Fusion::init() {
+    // initial estimate: E{ x(t0) }
+    x = 0;
+
+    // initial covariance: Var{ x(t0) }
+    P = 0;
+
+    mInitState = 0;
+    mCount[0] = 0;
+    mCount[1] = 0;
+    mCount[2] = 0;
+    mData = 0;
+}
+
+bool Fusion::hasEstimate() const {
+    return (mInitState == (MAG|ACC|GYRO));
+}
+
+bool Fusion::checkInitComplete(int what, const vec3_t& d) {
+    if (mInitState == (MAG|ACC|GYRO))
+        return true;
+
+    if (what == ACC) {
+        mData[0] += d * (1/length(d));
+        mCount[0]++;
+        mInitState |= ACC;
+    } else if (what == MAG) {
+        mData[1] += d * (1/length(d));
+        mCount[1]++;
+        mInitState |= MAG;
+    } else if (what == GYRO) {
+        mData[2] += d;
+        mCount[2]++;
+        if (mCount[2] == 64) {
+            // 64 samples is good enough to estimate the gyro drift and
+            // doesn't take too much time.
+            mInitState |= GYRO;
+        }
+    }
+
+    if (mInitState == (MAG|ACC|GYRO)) {
+        // Average all the values we collected so far
+        mData[0] *= 1.0f/mCount[0];
+        mData[1] *= 1.0f/mCount[1];
+        mData[2] *= 1.0f/mCount[2];
+
+        // calculate the MRPs from the data collection, this gives us
+        // a rough estimate of our initial state
+        mat33_t R;
+        vec3_t up(mData[0]);
+        vec3_t east(cross_product(mData[1], up));
+        east *= 1/length(east);
+        vec3_t north(cross_product(up, east));
+        R << east << north << up;
+        x[0] = matrixToMRPs(R);
+
+        // NOTE: we could try to use the average of the gyro data
+        // to estimate the initial bias, but this only works if
+        // the device is not moving. For now, we don't use that value
+        // and start with a bias of 0.
+        x[1] = 0;
+
+        // initial covariance
+        P = 0;
+    }
+
+    return false;
+}
+
+void Fusion::handleGyro(const vec3_t& w, float dT) {
+    const vec3_t wdT(w * dT);   // rad/s * s -> rad
+    if (!checkInitComplete(GYRO, wdT))
+        return;
+
+    predict(wdT);
+}
+
+status_t Fusion::handleAcc(const vec3_t& a) {
+    if (length(a) < 0.981f)
+        return BAD_VALUE;
+
+    if (!checkInitComplete(ACC, a))
+        return BAD_VALUE;
+
+    // ignore acceleration data if we're close to free-fall
+    const float l = 1/length(a);
+    update(a*l, Ba, accSTDEV*l);
+    return NO_ERROR;
+}
+
+status_t Fusion::handleMag(const vec3_t& m) {
+    // the geomagnetic-field should be between 30uT and 60uT
+    // reject obviously wrong magnetic-fields
+    if (length(m) > 100)
+        return BAD_VALUE;
+
+    if (!checkInitComplete(MAG, m))
+        return BAD_VALUE;
+
+    const vec3_t up( getRotationMatrix() * Ba );
+    const vec3_t east( cross_product(m, up) );
+    vec3_t north( cross_product(up, east) );
+
+    const float l = 1 / length(north);
+    north *= l;
+
+#if 0
+    // in practice the magnetic-field sensor is so wrong
+    // that there is no point trying to use it to constantly
+    // correct the gyro. instead, we use the mag-sensor only when
+    // the device points north (just to give us a reference).
+    // We're hoping that it'll actually point north, if it doesn't
+    // we'll be offset, but at least the instantaneous posture
+    // of the device will be correct.
+
+    const float cos_30 = 0.8660254f;
+    if (dot_product(north, Bm) < cos_30)
+        return BAD_VALUE;
+#endif
+
+    update(north, Bm, magSTDEV*l);
+    return NO_ERROR;
+}
+
+bool Fusion::checkState(const vec3_t& v) {
+    if (isnanf(length(v))) {
+        LOGW("9-axis fusion diverged. reseting state.");
+        P = 0;
+        x[1] = 0;
+        mInitState = 0;
+        mCount[0] = 0;
+        mCount[1] = 0;
+        mCount[2] = 0;
+        mData = 0;
+        return false;
+    }
+    return true;
+}
+
+vec3_t Fusion::getAttitude() const {
+    return x[0];
+}
+
+vec3_t Fusion::getBias() const {
+    return x[1];
+}
+
+mat33_t Fusion::getRotationMatrix() const {
+    return MRPsToMatrix(x[0]);
+}
+
+mat33_t Fusion::getF(const vec3_t& p) {
+    const float p0 = p.x;
+    const float p1 = p.y;
+    const float p2 = p.z;
+
+    // f(p, w)
+    const float p0p1 = p0*p1;
+    const float p0p2 = p0*p2;
+    const float p1p2 = p1*p2;
+    const float p0p0 = p0*p0;
+    const float p1p1 = p1*p1;
+    const float p2p2 = p2*p2;
+    const float pp = 0.5f * (1 - (p0p0 + p1p1 + p2p2));
+
+    mat33_t F;
+    F[0][0] = 0.5f*(p0p0 + pp);
+    F[0][1] = 0.5f*(p0p1 + p2);
+    F[0][2] = 0.5f*(p0p2 - p1);
+    F[1][0] = 0.5f*(p0p1 - p2);
+    F[1][1] = 0.5f*(p1p1 + pp);
+    F[1][2] = 0.5f*(p1p2 + p0);
+    F[2][0] = 0.5f*(p0p2 + p1);
+    F[2][1] = 0.5f*(p1p2 - p0);
+    F[2][2] = 0.5f*(p2p2 + pp);
+    return F;
+}
+
+mat33_t Fusion::getdFdp(const vec3_t& p, const vec3_t& we) {
+
+    // dF = | A = df/dp  -F |
+    //      |   0         0 |
+
+    mat33_t A;
+    A[0][0] = A[1][1] = A[2][2] = 0.5f * (p.x*we.x + p.y*we.y + p.z*we.z);
+    A[0][1] = 0.5f * (p.y*we.x - p.x*we.y - we.z);
+    A[0][2] = 0.5f * (p.z*we.x - p.x*we.z + we.y);
+    A[1][2] = 0.5f * (p.z*we.y - p.y*we.z - we.x);
+    A[1][0] = -A[0][1];
+    A[2][0] = -A[0][2];
+    A[2][1] = -A[1][2];
+    return A;
+}
+
+void Fusion::predict(const vec3_t& w) {
+    // f(p, w)
+    vec3_t& p(x[0]);
+
+    // There is a discontinuity at 2.pi, to avoid it we need to switch to
+    // the shadow of p when pT.p gets too big.
+    const float ptp(dot_product(p,p));
+    if (ptp >= 2.0f) {
+        p = -p * (1/ptp);
+    }
+
+    const mat33_t F(getF(p));
+
+    // compute w with the bias correction:
+    //  w_estimated = w - b_estimated
+    const vec3_t& b(x[1]);
+    const vec3_t we(w - b);
+
+    // prediction
+    const vec3_t dX(F*we);
+
+    if (!checkState(dX))
+        return;
+
+    p += dX;
+
+    const mat33_t A(getdFdp(p, we));
+
+    // G  = | G0  0 |  =  | -F  0 |
+    //      |  0  1 |     |  0  1 |
+
+    // P += A*P + P*At + F*Q*Ft
+    const mat33_t AP(A*transpose(P[0][0]));
+    const mat33_t PAt(P[0][0]*transpose(A));
+    const mat33_t FPSt(F*transpose(P[1][0]));
+    const mat33_t PSFt(P[1][0]*transpose(F));
+    const mat33_t FQFt(scaleCovariance(F, Q[0]));
+    P[0][0] += AP + PAt - FPSt - PSFt + FQFt;
+    P[1][0] += A*P[1][0] - F*P[1][1];
+    P[1][1] += Q[1];
+}
+
+void Fusion::update(const vec3_t& z, const vec3_t& Bi, float sigma) {
+    const vec3_t p(x[0]);
+    // measured vector in body space: h(p) = A(p)*Bi
+    const mat33_t A(MRPsToMatrix(p));
+    const vec3_t Bb(A*Bi);
+
+    // Sensitivity matrix H = dh(p)/dp
+    // H = [ L 0 ]
+    const float ptp(dot_product(p,p));
+    const mat33_t px(crossMatrix(p, 0.5f*(ptp-1)));
+    const mat33_t ppt(p*transpose(p));
+    const mat33_t L((8 / sqr(1+ptp))*crossMatrix(Bb, 0)*(ppt-px));
+
+    // update...
+    const mat33_t R(sigma*sigma);
+    const mat33_t S(scaleCovariance(L, P[0][0]) + R);
+    const mat33_t Si(invert(S));
+    const mat33_t LtSi(transpose(L)*Si);
+
+    vec<mat33_t, 2> K;
+    K[0] = P[0][0] * LtSi;
+    K[1] = transpose(P[1][0])*LtSi;
+
+    const vec3_t e(z - Bb);
+    const vec3_t K0e(K[0]*e);
+    const vec3_t K1e(K[1]*e);
+
+    if (!checkState(K0e))
+        return;
+
+    if (!checkState(K1e))
+        return;
+
+    x[0] += K0e;
+    x[1] += K1e;
+
+    // P -= K*H*P;
+    const mat33_t K0L(K[0] * L);
+    const mat33_t K1L(K[1] * L);
+    P[0][0] -= K0L*P[0][0];
+    P[1][1] -= K1L*P[1][0];
+    P[1][0] -= K0L*P[1][0];
+}
+
+// -----------------------------------------------------------------------
+
+}; // namespace android
+