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

00381 <span class="keyword">private</span>:
00382     <span class="keyword">mutable</span> FM::Vec hx;
00383 };
00384 
<a name="l00385"></a><a class="code" href="classBayesian__filter_1_1Linear__uncorrelated__observe__model.html">00385</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Linear__uncorrelated__observe__model.html">Linear_uncorrelated_observe_model</a> : <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Linrz__uncorrelated__observe__model.html">Linrz_uncorrelated_observe_model</a>
00386 <span class="comment">/* Linear observation model, uncorrelated observation noise</span>
00387 <span class="comment">    zp(k) = Hx(k) * x(k|k-1)</span>
00388 <span class="comment">    Enforces linear model invariant. Careful when deriving to to change this invariant!</span>
00389 <span class="comment"> */</span>
00390 {
00391 <span class="keyword">public</span>:
<a name="l00392"></a><a class="code" href="classBayesian__filter_1_1Linear__uncorrelated__observe__model.html#a0">00392</a>     <a class="code" href="classBayesian__filter_1_1Linear__uncorrelated__observe__model.html#a0">Linear_uncorrelated_observe_model</a> (size_t x_size, size_t z_size) :
00393         <a class="code" href="classBayesian__filter_1_1Linrz__uncorrelated__observe__model.html">Linrz_uncorrelated_observe_model</a>(x_size, z_size), hx(z_size)
00394     {}
<a name="l00395"></a><a class="code" href="classBayesian__filter_1_1Linear__uncorrelated__observe__model.html#a1">00395</a>     <span class="keyword">const</span> FM::Vec&amp; <a class="code" href="classBayesian__filter_1_1Linear__uncorrelated__observe__model.html#a1">h</a>(<span class="keyword">const</span> FM::Vec&amp; x)<span class="keyword"> const</span>
00396 <span class="keyword">    </span>{   <span class="comment">// Provide a linear implementation of functional h assumes model is already Linrz for Hx</span>
00397         hx.assign (FM::prod(<a class="code" href="classBayesian__filter_1_1Jacobian__observe__model.html#o0">Hx</a>,x));
00398         <span class="keywordflow">return</span> hx;
00399     }
00400 <span class="keyword">private</span>:
00401     <span class="keyword">mutable</span> FM::Vec hx;
00402 };
00403 
00404 
00405 <span class="comment">/*</span>
00406 <span class="comment"> * Bayesian Filter</span>
00407 <span class="comment"> *</span>
00408 <span class="comment"> * A Bayesian Filter uses Bayes rule to fuse the state probabilities</span>
00409 <span class="comment"> * of a prior and a likelhood function</span>
00410 <span class="comment"> */</span>
<a name="l00411"></a><a class="code" href="classBayesian__filter_1_1Bayes__filter__base.html">00411</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Bayes__filter__base.html">Bayes_filter_base</a> : <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Bayes__base.html">Bayes_base</a>
00412 {
00413     <span class="comment">// Empty</span>
00414 };
00415 
00416 <span class="comment">/*</span>
00417 <span class="comment"> * Likelihood Filter - Abstract filtering property</span>
00418 <span class="comment"> * Represents only the Bayesian Likelihood of a state observation</span>
00419 <span class="comment"> */</span>
<a name="l00420"></a><a class="code" href="classBayesian__filter_1_1Likelihood__filter.html">00420</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Likelihood__filter.html">Likelihood_filter</a> : <span class="keyword">virtual</span> <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Bayes__filter__base.html">Bayes_filter_base</a>
00421 {
00422 <span class="keyword">public</span>:
00423     <span class="comment">/* Virtual functions for filter algorithm */</span>
00424 
00425     <span class="keyword">virtual</span> <span class="keywordtype">void</span> <a class="code" href="classBayesian__filter_1_1Likelihood__filter.html#a0">observe</a> (<a class="code" href="classBayesian__filter_1_1Likelihood__observe__model.html">Likelihood_observe_model</a>&amp; h, <span class="keyword">const</span> FM::Vec&amp; z) = 0;
00426     <span class="comment">/* Observation state posterior using likelihood model h at z</span>
00427 <span class="comment">    */</span>
00428 };
00429 
00430 <span class="comment">/*</span>
00431 <span class="comment"> * Functional Filter - Abstract filtering property</span>
00432 <span class="comment"> * Represents only filter predict by a simple functional</span>
00433 <span class="comment"> * (non-stochastic) model</span>
00434 <span class="comment"> * </span>
00435 <span class="comment"> * A similar functional observe is not generally useful. The inverse of h is needed for observe!</span>
00436 <span class="comment"> */</span>
<a name="l00437"></a><a class="code" href="classBayesian__filter_1_1Functional__filter.html">00437</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Functional__filter.html">Functional_filter</a> : <span class="keyword">virtual</span> <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Bayes__filter__base.html">Bayes_filter_base</a>
00438 {
00439 <span class="keyword">public</span>:
00440     <span class="comment">/* Virtual functions for filter algorithm */</span>
00441 
00442     <span class="keyword">virtual</span> <span class="keywordtype">void</span> <a class="code" href="classBayesian__filter_1_1Functional__filter.html#a0">predict</a> (<a class="code" href="classBayesian__filter_1_1Functional__predict__model.html">Functional_predict_model</a>&amp; f) = 0;
00443     <span class="comment">/* Predict state with functional no noise model</span>
00444 <span class="comment">        Requires x(k|k), X(k|k) or internal equivilent</span>
00445 <span class="comment">        Predicts x(k+1|k), X(k+1|k), using predict model</span>
00446 <span class="comment">    */</span>
00447 };
00448 
00449 <span class="comment">/*</span>
00450 <span class="comment"> * State Filter - Abstract filtering property</span>
00451 <span class="comment"> * Represents only filter state and an update on that state</span>
00452 <span class="comment"> */</span>
<a name="l00453"></a><a class="code" href="classBayesian__filter_1_1State__filter.html">00453</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1State__filter.html">State_filter</a> : <span class="keyword">virtual</span> <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1Bayes__filter__base.html">Bayes_filter_base</a>
00454 {
00455 <span class="keyword">public</span>:
00456     <a class="code" href="classBayesian__filter_1_1State__filter.html#a0">State_filter</a> (size_t x_size);
00457     <span class="comment">/* Set constant sizes, state must not be empty (must be &gt;=1)</span>
00458 <span class="comment">        Exceptions:</span>
00459 <span class="comment">         bayes_filter_exception is x_size &lt; 1</span>
00460 <span class="comment">     */</span>
00461 
<a name="l00462"></a><a class="code" href="classBayesian__filter_1_1State__filter.html#o0">00462</a>     FM::Vec <a class="code" href="classBayesian__filter_1_1State__filter.html#o0">x</a>;          <span class="comment">// expected state</span>
00463 
00464     <span class="comment">/* Virtual functions for filter algorithm */</span>
00465 
00466     <span class="keyword">virtual</span> <span class="keywordtype">void</span> <a class="code" href="classBayesian__filter_1_1State__filter.html#a1">update</a> () = 0;
00467     <span class="comment">/* Update filters state</span>
00468 <span class="comment">        Updates x(k|k)</span>
00469 <span class="comment">    */</span>
00470 };
00471 
00472 
00473 <span class="comment">/*</span>
00474 <span class="comment"> * Kalman State Filter - Abstract filtering property</span>
00475 <span class="comment"> * Linear filter representation for 1st (mean) and 2nd (covariance) moments of a distribution</span>
00476 <span class="comment"> *</span>
00477 <span class="comment"> * Probability distributions are represted by state vector (x) and a covariance matix.(X)</span>
00478 <span class="comment"> *</span>
00479 <span class="comment"> * State (x) sizes is assumed to remain constant.</span>
00480 <span class="comment"> * The state and state covariance are public so they can be directly manipulated.</span>
00481 <span class="comment"> *  init: Should be called if x or X are altered</span>
00482 <span class="comment"> *  update: Guarantees that any internal changes made filter are reflected in x,X.</span>
00483 <span class="comment"> *  This allows considerable flexibility so filter implemtations can use different numerical representations</span>
00484 <span class="comment"> *</span>
00485 <span class="comment"> * Derived filters supply definititions for the abstract functions and determine the algorithm used</span>
00486 <span class="comment"> * to implement the filter.</span>
00487 <span class="comment"> */</span>
00488 
<a name="l00489"></a><a class="code" href="classBayesian__filter_1_1Kalman__state__filter.html">00489</a> <span class="keyword">class </span><a class="code" href="classBayesian__filter_1_1Kalman__state__filter.html">Kalman_state_filter</a> : <span class="keyword">public</span> <a class="code" href="classBayesian__filter_1_1State__filter.html">State_filter</a>
00490 {
00491 <span class="keyword">public</span>:
<a name="l00492"></a><a class="code" href="classBayesian__filter_1_1Kalman__state__filter.html#o0">00492</a>     FM::SymMatrix <a class="code" href="classBayesian__filter_1_1Kalman__state__filter.html#o0">X</a>;    <span class="comment">// state covariance</span>
00493 
00494     <a class="code" href="classBayesian__filter_1_1Kalman__state__filter.html#a0">Kalman_state_filter</a> (size_t x_size);
00495     <span class="comment">/* Initialise filter and set constant sizes</span>
00496 <span class="comment">     */</span>
00497 
00498     <span class="comment">/* Virtual functions for filter algorithm */</span>
00499 
00500     <span class="keyword">virtual</span> <span class="keywordtype">void</span> <a class="code" href="classBayesian__filter_1_1Kalman__state__filter.html#a1">init</a> () = 0;
00501     <span class="comment">/* Initialise from current state and state covariance</span>
00502 <span class="comment">        Requires x(k|k), X(k|k)</span>
00503 <span class="comment">    */</span>
00504     <span class="keywordtype">void</span> <a class="code" href="classBayesian__filter_1_1Kalman__state__filter.html#a2">init_kalman</a> (<span class="keyword">const</span> FM::Vec&amp; x, <span class="keyword">const</span> FM::SymMatrix&amp; X);
00505     <span class="comment">/* Initialise from a state and state covariance</span>
00506 <span class="comment">        Parameters that reference the instance's x and X members are valid</span>
00507 <span class="comment">    */</span>
00508     <span class="keyword">virtual</span> <span class="keywordtype">void</span> <a class="code" href="classBayesian__filter_1_1Kalman__state__filter.html#a3">update</a> () = 0;
00509     <span class="comment">/* Update filters state and state covariance </span>