bayesflt_8hpp-source.html

Bayes滤波器算法C++类说明文档,源码见Bayes滤波器算法

00252 
<a name="l00253"></a><a class="code" href="classBayesian__filter_1_1Likelihood__observe__model.html#a2">00253</a>     <span class="keyword">virtual</span> <span class="keywordtype">void</span> <a class="code" href="classBayesian__filter_1_1Likelihood__observe__model.html#a2">Lz</a> (<span class="keyword">const</span> FM::Vec&amp; zz)
00254     <span class="comment">// Set the observation zz about which to evaluate the Likelihood function</span>
00255     {
00256         <a class="code" href="classBayesian__filter_1_1Likelihood__observe__model.html#p0">z</a> = zz;
00257     }
00258 <span class="keyword">protected</span>:
<a name="l00259"></a><a class="code" href="classBayesian__filter_1_1Likelihood__observe__model.html#p0">00259</a>     FM::Vec <a class="code" href="classBayesian__filter_1_1Likelihood__observe__model.html#p0">z</a>;          <span class="comment">// z set by Lz</span>
00260 };
00261 
<a name="l00262"></a><a class="code" href="classBayesian__filter_1_1Functional__observe__model.html">00262</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Functional__observe__model.html">Functional_observe_model</a> : <span class="keyword">virtual</span> <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Observe__model__base.html">Observe_model_base</a>, <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Observe__function.html">Observe_function</a>
00263 <span class="comment">/* Functional (non-stochastic) observe model h</span>
00264 <span class="comment"> *  zp(k) = hx(x(k|k-1))</span>
00265 <span class="comment"> * This is a seperate fundamental model and not derived from likelihood because:</span>
00266 <span class="comment"> *  L is a delta function which isn't much use for numerical systems</span>
00267 <span class="comment"> * Defines an Interface without data members</span>
00268 <span class="comment"> */</span>
00269 {
00270 <span class="keyword">public</span>:
<a name="l00271"></a><a class="code" href="classBayesian__filter_1_1Functional__observe__model.html#a0">00271</a>     <a class="code" href="classBayesian__filter_1_1Functional__observe__model.html#a0">Functional_observe_model</a>(size_t <span class="comment">/*z_size*/</span>)
00272     {}
<a name="l00273"></a><a class="code" href="classBayesian__filter_1_1Functional__observe__model.html#a1">00273</a>     <span class="keyword">const</span> FM::Vec&amp; <a class="code" href="classBayesian__filter_1_1Functional__observe__model.html#a1">operator()</a>(<span class="keyword">const</span> FM::Vec&amp; x)<span class="keyword"> const</span>
00274 <span class="keyword">    </span>{   <span class="keywordflow">return</span> <a class="code" href="classBayesian__filter_1_1Observe__function.html#a0">h</a>(x);
00275     }
00276 
00277 };
00278 
<a name="l00279"></a><a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html">00279</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html">Parametised_observe_model</a> : <span class="keyword">virtual</span> <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Observe__model__base.html">Observe_model_base</a>, <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Observe__function.html">Observe_function</a>
00280 <span class="comment">/* Observation model parametised with a fixed z size</span>
00281 <span class="comment"> *  Includes the functional part of a noise model</span>
00282 <span class="comment"> *  Model is assume to have linear and non-linear components</span>
00283 <span class="comment"> *  Linear components need to be checked for conditioning</span>
00284 <span class="comment"> *  Non-linear components may be discontinous and need normalisation</span>
00285 <span class="comment"> */</span>
00286 {
00287 <span class="keyword">public</span>:
<a name="l00288"></a><a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html#a0">00288</a>     <a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html#a0">Parametised_observe_model</a>(size_t <span class="comment">/*z_size*/</span>)
00289     {}
00290     <span class="keyword">virtual</span> <span class="keyword">const</span> FM::Vec&amp; <a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html#a1">h</a>(<span class="keyword">const</span> FM::Vec&amp; x) <span class="keyword">const</span> = 0;
00291     <span class="comment">// Functional part of addative model</span>
<a name="l00292"></a><a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html#a2">00292</a>     <span class="keyword">virtual</span> <span class="keywordtype">void</span> <a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html#a2">normalise</a> (FM::Vec&amp; <span class="comment">/*z_denorm*/</span>, <span class="keyword">const</span> FM::Vec&amp; <span class="comment">/*z_from*/</span>) <span class="keyword">const</span>
00293     <span class="comment">// Discontinous h. Normalise one observation (z_denorm) from another</span>
00294     {}  <span class="comment">//  Default normalistion, z_denorm unchanged</span>
00295     
<a name="l00296"></a><a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html#o0">00296</a>     <a class="code" href="classBayesian__filter_1_1Numerical__rcond.html">Numerical_rcond</a> <a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html#o0">rclimit</a>;
00297     <span class="comment">// Reciprocal condition number limit of linear components when factorised or inverted</span>
00298 };
00299 
<a name="l00300"></a><a class="code" href="classBayesian__filter_1_1Uncorrelated__addative__observe__model.html">00300</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Uncorrelated__addative__observe__model.html">Uncorrelated_addative_observe_model</a> : <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html">Parametised_observe_model</a>
00301 <span class="comment">/* Observation model, uncorrelated addative observation noise</span>
00302 <span class="comment">    Z(k) = I * Zv(k) observe noise variance vector Zv</span>
00303 <span class="comment"> */</span>
00304 {
00305 <span class="keyword">public</span>:
<a name="l00306"></a><a class="code" href="classBayesian__filter_1_1Uncorrelated__addative__observe__model.html#a0">00306</a>     <a class="code" href="classBayesian__filter_1_1Uncorrelated__addative__observe__model.html#a0">Uncorrelated_addative_observe_model</a> (size_t z_size) :
00307         <a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html">Parametised_observe_model</a>(z_size), <a class="code" href="classBayesian__filter_1_1Uncorrelated__addative__observe__model.html#o0">Zv</a>(z_size)
00308     {}
<a name="l00309"></a><a class="code" href="classBayesian__filter_1_1Uncorrelated__addative__observe__model.html#o0">00309</a>     FM::Vec <a class="code" href="classBayesian__filter_1_1Uncorrelated__addative__observe__model.html#o0">Zv</a>;         <span class="comment">// Noise Variance</span>
00310 };
00311 
<a name="l00312"></a><a class="code" href="classBayesian__filter_1_1Correlated__addative__observe__model.html">00312</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Correlated__addative__observe__model.html">Correlated_addative_observe_model</a> : <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html">Parametised_observe_model</a>
00313 <span class="comment">/* Observation model, correlated addative observation noise</span>
00314 <span class="comment">    Z(k) = observe noise covariance</span>
00315 <span class="comment"> */</span>
00316 {
00317 <span class="keyword">public</span>:
<a name="l00318"></a><a class="code" href="classBayesian__filter_1_1Correlated__addative__observe__model.html#a0">00318</a>     <a class="code" href="classBayesian__filter_1_1Correlated__addative__observe__model.html#a0">Correlated_addative_observe_model</a> (size_t z_size) :
00319         <a class="code" href="classBayesian__filter_1_1Parametised__observe__model.html">Parametised_observe_model</a>(z_size), <a class="code" href="classBayesian__filter_1_1Correlated__addative__observe__model.html#o0">Z</a>(z_size,z_size)
00320     {}
<a name="l00321"></a><a class="code" href="classBayesian__filter_1_1Correlated__addative__observe__model.html#o0">00321</a>     FM::SymMatrix <a class="code" href="classBayesian__filter_1_1Correlated__addative__observe__model.html#o0">Z</a>;    <span class="comment">// Noise Covariance (not necessarly dense)</span>
00322 };
00323 
<a name="l00324"></a><a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html">00324</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html">Jacobian_observe_model</a> : <span class="keyword">virtual</span> <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Observe__model__base.html">Observe_model_base</a>
00325 <span class="comment">/* Linrz observation model Hx, h about state x (fixed size)</span>
00326 <span class="comment">    Hx(x(k|k-1) = Jacobian of h with respect to state x</span>
00327 <span class="comment">    Normalisation consistency Hx: Assume normalise will be from h(x(k|k-1)) so result is consistent with Hx</span>
00328 <span class="comment"> */</span>
00329 {
00330 <span class="keyword">public</span>:
<a name="l00331"></a><a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html#o0">00331</a>     FM::Matrix <a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html#o0">Hx</a>;      <span class="comment">// Model</span>
00332 <span class="keyword">protected</span>: <span class="comment">// Jacobian model is not sufficient, it is used to build Linrz observe model's</span>
<a name="l00333"></a><a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html#b0">00333</a>     <a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html#b0">Jacobian_observe_model</a> (size_t x_size, size_t z_size) :
00334         <a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html#o0">Hx</a>(z_size, x_size)
00335     {}
00336 };
00337 
<a name="l00338"></a><a class="code" href="classBayesian__filter_1_1Linrz__correlated__observe__model.html">00338</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Linrz__correlated__observe__model.html">Linrz_correlated_observe_model</a> : <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Correlated__addative__observe__model.html">Correlated_addative_observe_model</a>, <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html">Jacobian_observe_model</a>
00339 <span class="comment">/* Linrz observation model Hx, h with repespect to state x (fixed size)</span>
00340 <span class="comment">    correlated observation noise</span>
00341 <span class="comment">    zp(k) = h(x(k-1|k-1)</span>
00342 <span class="comment">    Hx(x(k|k-1) = Jacobian of f with respect to state x</span>
00343 <span class="comment">    Z(k) = observe noise covariance</span>
00344 <span class="comment"> */</span>
00345 {
00346 <span class="keyword">public</span>:
<a name="l00347"></a><a class="code" href="classBayesian__filter_1_1Linrz__correlated__observe__model.html#a0">00347</a>     <a class="code" href="classBayesian__filter_1_1Linrz__correlated__observe__model.html#a0">Linrz_correlated_observe_model</a> (size_t x_size, size_t z_size) :
00348         <a class="code" href="classBayesian__filter_1_1Correlated__addative__observe__model.html">Correlated_addative_observe_model</a>(z_size), <a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html">Jacobian_observe_model</a>(x_size, z_size)
00349     {}
00350 };
00351 
<a name="l00352"></a><a class="code" href="classBayesian__filter_1_1Linrz__uncorrelated__observe__model.html">00352</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Linrz__uncorrelated__observe__model.html">Linrz_uncorrelated_observe_model</a> : <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Uncorrelated__addative__observe__model.html">Uncorrelated_addative_observe_model</a>, <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html">Jacobian_observe_model</a>
00353 <span class="comment">/* Linrz observation model Hx, h with repespect to state x (fixed size)</span>
00354 <span class="comment">    uncorrelated observation noise</span>
00355 <span class="comment">    zp(k) = h(x(k-1|k-1)</span>
00356 <span class="comment">    Hx(x(k|k-1) = Jacobian of f with respect to state x</span>
00357 <span class="comment">    Zv(k) = observe noise covariance</span>
00358 <span class="comment"> */</span>
00359 {
00360 <span class="keyword">public</span>:
<a name="l00361"></a><a class="code" href="classBayesian__filter_1_1Linrz__uncorrelated__observe__model.html#a0">00361</a>     <a class="code" href="classBayesian__filter_1_1Linrz__uncorrelated__observe__model.html#a0">Linrz_uncorrelated_observe_model</a> (size_t x_size, size_t z_size) :
00362         <a class="code" href="classBayesian__filter_1_1Uncorrelated__addative__observe__model.html">Uncorrelated_addative_observe_model</a>(z_size), <a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html">Jacobian_observe_model</a>(x_size, z_size)
00363     {}
00364 };
00365 
<a name="l00366"></a><a class="code" href="classBayesian__filter_1_1Linear__correlated__observe__model.html">00366</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Linear__correlated__observe__model.html">Linear_correlated_observe_model</a> : <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Linrz__correlated__observe__model.html">Linrz_correlated_observe_model</a>
00367 <span class="comment">/* Linear observation model, correlated observation noise</span>
00368 <span class="comment">    zp(k) = Hx(k) * x(k|k-1)</span>
00369 <span class="comment">    Enforces linear model invariant. Careful when deriving to to change this invariant!</span>
00370 <span class="comment"> */</span>
00371 {
00372 <span class="keyword">public</span>:
<a name="l00373"></a><a class="code" href="classBayesian__filter_1_1Linear__correlated__observe__model.html#a0">00373</a>     <a class="code" href="classBayesian__filter_1_1Linear__correlated__observe__model.html#a0">Linear_correlated_observe_model</a> (size_t x_size, size_t z_size) :
00374         <a class="code" href="classBayesian__filter_1_1Linrz__correlated__observe__model.html">Linrz_correlated_observe_model</a>(x_size, z_size), hx(z_size)
00375     {}
<a name="l00376"></a><a class="code" href="classBayesian__filter_1_1Linear__correlated__observe__model.html#a1">00376</a>     <span class="keyword">const</span> FM::Vec&amp; <a class="code" href="classBayesian__filter_1_1Linear__correlated__observe__model.html#a1">h</a>(<span class="keyword">const</span> FM::Vec&amp; x)<span class="keyword"> const</span>
00377 <span class="keyword">    </span>{   <span class="comment">// Provide a linear implementation of functional h assumes model is already Linrz for Hx</span>
00378         hx.assign (FM::prod(<a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html#o0">Hx</a>,x));
00379         <span class="keywordflow">return</span> hx;
00380     }