summaryrefslogtreecommitdiff log msg author committer range
path: root/KalmanIntro.mdwn
blob: b9b2b52f614fc45d16cc46a7b9f3fe63eb7dd6a4 (plain)
 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719  # (Yet Another) Introduction to the Kalman Filter [[(See also our mathematical notation page)|MathSymbols]] **Introduction to the Kalman Filter** 1. [Basics](#ChapBasics) 2. Example 3. [Derivation](#ChapDerivation) 4. Complications 5.
Todo
---- ## Basics Given a system model, an initial system state, and a sequence of noisy measurements, a Kalman filter can be constructed to produce a sequence of state estimates that are optimal in the sense that they minimize the expected square-error between the estimates and the true system state. **Summary of Kalman Filter Symbols**

Symbol

Dimension

Name

x

n×1

system state vector

Φ (Phi)

n×n

state transition matrix

P

n×n

state error covariance matrix

u

u×1

control input vector

Γ (Gamma)

n×u

control input matrix

w

q×1

process noise vector

Λ (Lambda)

n×q

process noise input matrix

Q

q×q

process noise covariance matrix

y

m×1

output vector

H

m×n

output matrix

ν (nu)

m×1

measurement noise vector

R

m×m

measurement noise covariance matrix

K

n×m

Kalman gain, feedback matrix

Table (1)

**Typical Kalman Filter Equations**

Equation

Description

$\bf{\hat x_0} = \it E[\bf{x_0}]$

state initialization

$\bf{P_0} = \it E[(\bf{\hat x_0 - x_0})^2]$

state error initialization

$\bf{x_k^-} = \bf{\Phi_{k-1} \hat x_{k-1} + \Gamma_{k-1} u_{k-1}}$

state propagation
Previous knowledge of the system is used to guess the future state.

$\bf{P_k-} = \bf{\Phi_{k-1} P_{k-1} \Phi_{k-1}T + \Lambda_{k-1} Q_{k-1} \Lambda_{k-1}^T}$

state error propagation
System dynamics effect the error. Process noise increases the error.

$\bf{K_k} = \bf{P_k-\, H_kT (H_k P_k-\, H_kT + R_k)^{-1}}$

Kalman gain update
Recursive least squares solution, trades off state uncertainty and measurement noise.

$\bf{\hat x_k} = \bf{x_k- + K_k (\tilde y_k - H_k x_k-)}$

measurement update
New optimal state estimate

$\bf{P_k} = \bf{((P_k-){-1} + H_k R_k{-1} H_k){-1}}$

state error update
new measurements add to knowledge of system, estimated error decreases

Table (2)

Start with the discreet linear dynamic system equation #!latex (-) $$\displaystyle (1.1)\quad \bf{x_{k+1}} = \bf{\Phi_k\ x_k} + \bf{\Gamma_k} \bf{u_k} + \bf{\Lambda_k} \bf{w_k}$$ And the measurement equation #!latex (-) $$\displaystyle (1.2)\quad \bf{\tilde y_k} = \bf{H_k} \bf{x_k} + \bf{\nu_k}$$ Where the subscript (·k) indicates the value at time (tk) Although the form of the system equations would seem to apply only to strictly linear systems, as usual, various linearizations and quasi-non-linear techniques can be applied to extend the Kalman filter to a variety of real world non-linear problems. Making some key assumptions will greatly simplify what follows.
Assume that - For each time step (k); (Φk), (Γk), (Λk), (**H**k) and (**u**k) are known without error - Noise components (**w**k) and (νk) are uncorrelated Gaussian random sequences with zero mean - The noise covariance matrices are known, (cov[**w**k]=**Q**k), (cov[νk]=**R**k) In principle these assumptions can be relaxed somewhat. For example, if any of the system matrices contain random components, those components may be factored out into the noise vectors. Similarly, any noise process that is representable as a linear combination of white Gaussian noise can be modeled by adding extra states to the system and driving those extra states with zero mean Gaussian noise. Even if the noise matrices are not known exactly, they can be estimated, or calculated on line, or discovered by system tuning. The exact quantity minimized by a Kalman filter is the matrix (**P**), the _state error covariance matrix_ #!latex (-) $$\displaystyle (1.3)\quad \bf{P} \equiv \it{E}[\bf{(\hat x-x)}^2]$$ Where (_E_ ) is a function returning the expectation of a random variable The estimated state ($\\bf\{\\hat x\}$) that minimizes (**P**) is found by solving a weighted least squares problem. The weights for the computation ultimately come from the process noise covariance matrix (**Q**) and the measurement noise covariance matrix (**R**). The actual output of the filter is a sequence of linear combinations formed from the predicted state (**x**\_), and the current measurements of the system ($\\bf\{\\tilde y\}$). Technically these outputs are known as the _best linear unbiased estimates_. The existence of such estimates is guaranteed only with certain restrictions, see the section on complications. (FIXME: complications section not built yet) Typically the Kalman filter is implemented recursively as a _predictor-corrector_ system. In the prediction step, knowledge of the past state of the system is used to extrapolate to the system's state at the next time step. Once the next time step is actually reached and new measurements become available, the new measurements are used to refine the filter's estimate of the true state like this #!latex (-) $$\displaystyle (1.4)\quad \bf{\hat x_k} = \bf{x_k^-} + \bf{K_k}(\bf{\tilde y_k}-\bf{H_k}\bf{x_k^-})$$ Where the matrix (**K**k), the _Kalman gain_, or _feedback matrix_ , is chosen so that (**P**k) is minimized Later there will be some [justification](#EqNum3Dot14), but for now accept that one possible choice for (**K**k) is #!latex (-) $$\displaystyle (1.5)\quad \bf{K_k} = \bf{P_k^-\, H_k^T (H_k P_k^-\, H_k^T + R_k)^{-1}} = \bf{\frac {P_k^-\, H_k^T}{H_k P_k^-\, H_k^T + R_k}}$$ Where the _measurement noise covariance matrix_ (**R**k) is defined as #!latex (-) $$\displaystyle (1.6) \quad \bf{R} \equiv \mbox{\rm cov}[\bf{\nu}] = \it E[\bf{\nu}^2]$$ The relative magnitudes of matrices (**R**k) and (**P**k) control a trade-off between the filter's use of predicted state estimate (**x**k\_) and measurement ($\\bf\{\\tilde y\_k\}$). Consider some limits on (**K**k) #!latex (-) $$\displaystyle (1.7a)\quad \lim\limits_{\bf{R_k \to 0}} \bf{{P_k^-\, H_k^T} \over {H_k P_k^-\, H_k^T + R_k}} = \bf{H_k^{-1}}$$ #!latex (-) $$\displaystyle (1.7b)\quad \lim\limits_{\bf{P_k \to 0}} \bf{{P_k^-\, H_k^T} \over {H_k P_k^-\, H_k^T + R_k}} = \bf 0$$ Substituting the first limit into the measurement update equation [(1.4)](#EqNum1Dot4) suggests that when the magnitude of (**R**) is small, meaning that the measurements are accurate, the state estimate depends mostly on the measurements. Likewise when the state is known accurately, then (**H P**- **H**T) is small compared to (**R**), and the filter mostly ignores the measurements relying instead on the prediction derived from the previous state (**x**k\_). ---- ## Example (FIXME: no example yet) ---- ## Derivation The desired Kalman filter implementation is linear and recursive, that is, given the old filter state ($\{\\cal F\}\_\{k-1\}$) and new measurements ($\{\\cal M\}\_k$) the the new filter state is calculated as #!latex (-) $$\displaystyle (3.1a)\quad {\cal F}_k^- = \it g [{\cal F}_{k-1}]$$ #!latex (-) $$\displaystyle (3.1b)\quad {\cal F}_k = \it f [{\cal F}_k^-\,,\ {\cal M}_k]$$ Where ( _f_ and _g_ ) are linear functions Referring to [table(2)](#TblNum2), the sequence of equations shown are both linear and recursive as desired. Though so far, little explanation has been given to justify their use. In order to maintain some credibility, and to prove we aren't total mathematical wimps, we will attempt a derivation of these equations. The goal here is not mathematical perfection, rather clarity without serious error. Fundamentally Kalman filtering involves solving a least squares error problem. Specifically, given a state estimate (**x**\_) of known covariance (**P**-) and a new set of measurements ($\\bf\{\\tilde y\}$) with covariance (**R**), find a new state estimate ($\\bf\{\\hat x\}$) such that the new state error covariance (**P**) is minimized. For the predicted state error write #!latex (-) $$\displaystyle (3.2) \quad \bf \mbox{\rm cov}[\bf{x^- - x}] = \bf P^-$$ Likewise the measurement error is #!latex (-) $$\displaystyle (3.3) \quad \bf \mbox{\rm cov}[\bf{\tilde y - y}] = \bf R$$ Seek a new state vector ($\\bf\{\\hat x\}$) such that the (n+m × 1) combined state error vector #!latex (-) $$\displaystyle (3.4)\quad \pmatrix{ \bf I \cr \bf H } \bf{\hat x} - \pmatrix{ \bf x^- \cr \bf{\tilde y} }$$ Is minimized in the weighted least squares sense The chief difficulty with Kalman filter theory is contained in [(3.4)](#EqNum3Dot4). The errors in [(3.2)](#EqNum3Dot2) and [(3.3)](#EqNum3Dot3) are defined in terms of the true state and exact measurements, but while operating the filter, all that is available are estimates and noisy measurements. In equation [(3.4)](#EqNum3Dot4) the measurement ($\\bf\{\\tilde y\}$) and the estimated state (**x**\_) have been used where the exact values would seem to be appropriate. Here's a simple-minded explanation of why it's ok to use the imprecise values in [(3.4)](#EqNum3Dot4): The noise has been assumed to be zero-mean, therefore the _expectation_ of the approximate values are the correct true values, so using the approximate values will produce unbiased estimates. Further, since it has been assumed that the covariance of both ($\\bf\{\\tilde y\}$) and (**x**\_) are known, the weighted least squares process has enough information to properly minimize the total square error. Whether the previous paragraph constitutes a satisfactory justification of [(3.4)](#EqNum3Dot4) is a question worth pondering. If not there are many, many, many, books that address this issue (try looking up _stochastic least squares_). (However the chances of finding a very clear explanation appear to be low.) The issue is significant in as much as once [(3.4)](#EqNum3Dot4) is accepted, Kalman filtering as a theoretical matter is essentially solved. Though, solved in this sense is still far from a practical implementation. To derive a practical filter, some clever arithmetic is required. A weighted least squares solution minimizes a scalar cost function ($\\cal\{J\})$, here written in matrix form #!latex (-) $$\displaystyle (3.5)\quad \cal{J}[\bf \epsilon , \bf W] = \bf{\epsilon}^T \, \bf W \, \bf \epsilon$$ Where the matrix (**W**) is the weighting for the cost function. The weighting can be any positive definite matrix. In formulating the Kalman filter the noise statistics have been defined in terms of covariance, and since elements with large covariance should be given low weight, the correct weighting matrix for this problem is the inverse of the covariance matrix. This is appealing intuitively and not too difficult to prove, though no proof is given here. Using the inverse covariance as the weighting function and referring to [(3.4)](#EqNum3Dot4) the cost function for the Kalman filter is #!latex (-) $$\displaystyle (3.6)\quad \cal{J} = \pmatrix{ \bf{I\, \hat x_k} - \bf{x_k^-} \cr \bf{H_k \hat x_k} - \bf{\tilde y_k} }^{\rm T} \pmatrix{ \bf P_k^- & \bf 0 \cr \bf 0 & \bf R_k }^{\rm -1} \pmatrix{ \bf{I \, \hat x_k} - \bf{x_k^-} \cr \bf{H_k \hat x_k} - \bf{\tilde y_k} }$$ Where the block matrices shown have dimensions (n+m × 1) and (n+m × n+m) respectively Using the mapping suggested by [(3.4)](#EqNum3Dot4) #!latex (-) $$\displaystyle (3.7)\quad \left\{ \pmatrix{\bf I \cr \bf H_k} \rightarrow \bf H,\ \bf{\hat x_k} \rightarrow \bf{\hat x},\ \pmatrix{\bf x_k^- \cr \bf{\tilde y_k}} \rightarrow \bf{\tilde y},\ \pmatrix{\bf P_k^- & 0 \cr 0 & \bf R_k} \rightarrow \bf{R} \right\}$$ The cost function can be stated in a slightly more generic form as #!latex (-) \begin{eqnarray*} (3.8)\quad \vphantom{\frac{0}{0}} \cal{J} &=& \bf{(H\,\hat x - \tilde y)^T\,R^{-1} (H\,\hat x - \tilde y})\cr \vphantom{\frac{0}{0}} &=& (\bf{\hat x^T \, H^T - \tilde y^T})\,\bf R^{-1} (\bf{H\,\hat x - \tilde y})\cr \vphantom{\frac{0}{0}} &=& (\bf{\hat x^T\,H^T - \tilde y^T}) (\bf{R^{-1}\,H\,\hat x - R^{-1}\,\tilde y})\cr \vphantom{\frac{0}{0}} &=& \bf{\hat x^T\,H^T\,R^{-1}\,H\,\hat x - \hat x^T\,H^T\,R^{-1}\,\tilde y - \tilde y^T\,R^{-1}\,H\,\hat x + \tilde y^T\,R^{-1}\,\tilde y} \end{eqnarray*} In [(3.8)](#EqNum3Dot8) the cost function has been expanded in anticipation of taking derivatives. Note that a matrix derivative typically involves taking a transpose. Returning to the generic minimization problem, take the partial derivative #!latex (-) \begin{eqnarray*} (3.9)\quad \vphantom{\frac{0}{0}} \frac{\partial\cal{J}} {\partial\bf{\hat x}} &=& \bf{H^T\,R^{-1}\,H\,\hat x + (\hat x^T\,H^T\,R^{-1}\,H)^T - H^T\,R^{-1}\,\tilde y - (\tilde y^T\,R^{-1}\,H)^T}\cr \vphantom{\frac{0}{0}} &=& 2\ \bf{H^T\,R^{-1}\,H\,\hat x} - 2\ \bf{H^T\,R^{-1}\,\tilde y}\cr \end{eqnarray*} Set the derivative to zero and solve for ($\\bf\{\\hat x\}$) #!latex (-) $$\displaystyle (3.10)\quad \bf{\hat x} = \bf{(H^T\,R^{-1}\,H)^{-1}\,H^T\,R^{-1}\,\tilde y}$$ This is the classic weighted least squares solution. Mapping [(3.10)](#EqNum3Dot10) back to our original problem using [(3.7)](#EqNum3Dot7) gets #!latex (-) \begin{eqnarray*} (3.11)\quad \vphantom{\frac{0}{0}} \bf{\hat x_k} &=& \left( \pmatrix{\bf I \cr \bf H_k}^T \pmatrix{\bf P_k^- & \bf 0 \cr \bf 0 & \bf R_k}^{-1} \pmatrix{\bf I \cr \bf H_k} \right)^{-1} \pmatrix{\bf I \cr \bf H_k}^T \pmatrix{\bf P_k^- & \bf 0 \cr \bf 0 & \bf R_k}^{-1} \pmatrix{\bf x_k^- \cr \bf{\tilde y_k}}\cr \vphantom{\frac{0}{0}} &=& (\bf{P_k^-}^{-1}+\bf{H_k^T\,R_k^{-1}\,H_k})^{-1} (\bf{P_k^-}^{-1}\,\bf{x_k^- + H_k^T\,R_k^{-1}\,\tilde y_k}) \end{eqnarray*} There's nothing wrong with the form given in [(3.11)](#EqNum3Dot11), but for various reasons it's desirable to exhibit the measurement update form [(1.4)](#EqNum1Dot4). To do this everyone uses the _matrix inversion lemma_ #!latex (-) $$\displaystyle (3.12)\quad \bf{(A-B^T\,C^{-1}\,B)^{-1} = A^{-1}-A^{-1}\,B^T(B\,A^{-1}B^T-C)^{-1}B\,A^{-1}}$$ Using the mapping #!latex (-) $$\displaystyle (3.13)\quad \left\{ \bf A \rightarrow \bf{P_k^-}^{-1},\ \ \bf B \rightarrow \bf H_k ,\ \ \bf C \rightarrow -\bf R_k \ \ \right\}\$$ The solution given by [(3.11)](#EqNum3Dot11) becomes #!latex (-) \begin{eqnarray*} (3.14)\quad \vphantom{\frac{0}{0}} \bf{\hat x_k} &=& \bf{ (P_k^- - P_k^-\,H_k^T(H_k\,P_k^-\,H_k^T + R_k)^{-1}H_k\,P_k^-) ({P_k^-}^{-1}\,x_k^- + H_k^T\,R_k^{-1}\,\tilde y_k) }\cr \vphantom{\frac{0}{0}} &=&\bf{ P_k^-\,{P_k^-}^{-1}\,x_k^- + P_k^-\,H_k^T\,R_k^{-1}\,\tilde y_k - P_k^-\,H_k^T(H_k\,P_k^-\,H_k^T + R_k)^{-1}H_k\,P_k^-\,{P_k^-}^{-1}\,x_k^- - }\cr \vphantom{\frac{0}{0}} & &\bf{ P_k^-\,H_k^T(H_k\,P_k^-\,H_k^T + R_k)^{-1}H_k\,P_k^-\,H_k^T\,R_k^{-1}\,\tilde y_k }\cr \vphantom{\frac{0}{0}} &=&\bf{ x_k^- - P_k^-\,H_k^T(H_k\,P_k^-\,H_k^T + R_k)^{-1}H_k\,x_k^- - }\cr \vphantom{\frac{0}{0}} & &\bf{ P_k^-\,H_k^T(I-(H_k\,P_k^-\,H_k^T + R_k)^{-1}H_k\,P_k^-\,H_k^T)R_k^{-1}\,\tilde y_k }\cr \vphantom{\frac{0}{0}} &=&\bf{ x_k^- - P_k^-\,H_k^T(H_k\,P_k^-\,H_k^T + R_k)^{-1}H_k\,x_k^-\ - P_k^-\,H_k^T \cdot }\cr \vphantom{\frac{0}{0}} & &\bf{ \left( (H_k\,P_k^-\,H_k^T + R_k)^{-1}(H_k\,P_k^-\,H_k^T\ + R_k) - (H_k\,P_k^-\,H_k^T + R_k)^{-1}H_k\,P_k^-\,H_k^T \right) \cdot }\cr \vphantom{\frac{0}{0}} & &\bf{R_k^{-1}\,\tilde y_k}\cr \vphantom{\frac{0}{0}} &=&\bf{ x_k^- - P_k^-\,H_k^T(H_k\,P_k^-\,H_k^T + R_k)^{-1}H_k\,x_k^- - }\cr \vphantom{\frac{0}{0}} & &\bf{ P_k^-\,H_k^T(H_k\,P_k^-\,H_k^T + R_k)^{-1} (H_k\,P_k^-\,H_k^T + R_k - H_k\,P_k^-\,H_k^T)R_k^{-1}\,\tilde y_k }\cr \vphantom{\frac{0}{0}} &=&\bf{ x_k^- - P_k^-\,H_k^T(H_k\,P_k^-\,H_k^T + R_k)^{-1}H_k\,x_k^- - }\cr \vphantom{\frac{0}{0}} & &\bf{ P_k^-\,H_k^T(H_k\,P_k^-\,H_k^T + R_k)^{-1}R_k\,R_k^{-1}\,\tilde y_k }\cr \vphantom{\frac{0}{0}} &=&\bf{ x_k^- - P_k^-\,H_k^T(H_k\,P_k^-\,H_k^T + R_k)^{-1}(H_k\,x_k^- - \tilde y_k) }\cr \vphantom{\frac{0}{0}} &=&\bf{ x_k^- + K_k (\tilde y_k - H_k\,x_k^-) }\cr \end{eqnarray*} Which matches the measurement update form given in [(1.4)](#EqNum1Dot4) with (**K**k) as in [(1.5)](#EqNum1Dot5) It's possible to derive the state error update equation directly from the definition given in [(1.3)](#EqNum1Dot3) and repeated here #!latex (-) $$\displaystyle (3.15)\quad \bf{P} \equiv \it{E}[\bf{(\hat x-x)}^2]$$ It would be very convenient to write the true state (**x**k) in terms of ($\\bf\{\\hat x\_k\}$), in fact the previous hand-waving argument in support of equation [(3.4)](#EqNum3Dot4) implies this must be possible. Using the mapping defined in [(3.7)](#EqNum3Dot7), the measurement equation is #!latex (-) $$\displaystyle (3.16)\quad \bf{\tilde y = H\,x + \pmatrix{\bf{x_k^- - x_k} \cr \bf \nu}}$$ According to our least squares solution [(3.10)](#EqNum3Dot10) #!latex (-) \begin{eqnarray*} \vphantom{\frac{0}{0}} (3.17)\quad \bf{\hat x} &=& \bf{(H^T\,R^{-1}\,H)^{-1} H^T\,R^{-1}\,\tilde y }\cr \vphantom{\frac{0}{0}} &=& \bf{(H^T\,R^{-1}\,H)^{-1}H^T\,R^{-1}} \left( \bf{H\,x + \pmatrix{\bf{x_k^- - x_k} \cr \bf \nu }} \right)\cr \vphantom{\frac{0}{0}} &=& \bf{x + (H^T\,R^{-1}\,H)^{-1} H^T\,R^{-1}\pmatrix{\bf{x_k^- - x_k} \cr \bf \nu }}\cr \end{eqnarray*} Putting this into the definition for (**P**) #!latex (-) \begin{eqnarray*} (3.18)\quad \vphantom{\frac{0}{0}} \bf P &=& \it{E}[(\bf{\hat x - x})^2] = \it{E} \left[ \left(\bf{(H^T\,R^{-1}\,H)^{-1} H^T\,R^{-1} \pmatrix{\bf{x_k^- - x_k} \cr \bf \nu }}\right)^2 \right]\cr \vphantom{\frac{0}{0}} &=&\left( \bf{(H^T\,R^{-1}\,H)^{-1}H^T\,R^{-1}} \right) \it{E} \left[ \pmatrix{\bf{x_k^- - x_k} \cr \bf \nu }^2 \right] \left( \bf{(H^T\,R^{-1}\,H)^{-1}H^T\,R^{-1}} \right)^{\bf T}\cr \vphantom{\frac{0}{0}} &=&\bf{ \left( (H^T\,R^{-1}\,H)^{-1}H^T\,R^{-1} \right) \,R\, \left( (H^T\,R^{-1}\,H)^{-1}H^T\,R^{-1} \right)^T }\cr \vphantom{\frac{0}{0}} &=& (\bf{H^T\,R^{-1} H})^{-1}\cr \end{eqnarray*} Substituting back the values for (**H**) and (**R**) from [(3.7)](#EqNum3Dot7) #!latex (-) \begin{eqnarray*} (3.19)\quad \vphantom{\frac{0}{0}} \bf P_k &=& \bf{ \left( \pmatrix{\bf I \cr \bf H_k}^T \pmatrix{\bf P_k^- & \bf 0 \cr \bf 0 & \bf R_k}^{-1} \pmatrix{\bf I \cr \bf H_k} \right)^{-1} }\cr \vphantom{\frac{0}{0}} &=& (\bf{{P_k^-}^{-1} + H_k^T\,R_k^{-1} H_k})^{-1}\cr \end{eqnarray*} Equation [(3.19)](#EqNum3Dot19) performs the "measurement update" for the state error estimate (**P**). It can be written as a parallel sum #!latex (-) $$\displaystyle (3.20)\quad \bf{ {1 \over P_k} = {1 \over P_k^-} + H_k^T {1 \over R_k} H_k }$$ The last term in [(3.20)](#EqNum3Dot20) can be considered to represent the extra information added to the state because of the current measurement. The form of [(3.20)](#EqNum3Dot20) makes sense because adding new information (positive (**R**k-1)) will tend to decrease the system uncertainty. Now all that remains is to derive the state error propagation equation, again using definition [(1.3)](#EqNum1Dot3) #!latex (-) $$\displaystyle (3.21)\quad \bf P_{k+1}^- = \it{E} [ ( \bf{ x_{k+1}^- - x_{k+1} } )^2 ]$$ Applying the state propagation equation [(1.1)](#EqNum1Dot1) to the inside of [(3.21)](#EqNum3Dot21), the control input, being common, cancels out #!latex (-) $$\displaystyle (3.22)\quad ( \bf{ x_{k+1}^- - x_{k+1} } ) = \bf{ \Phi_k (\hat x_k - x_k) + \Gamma_k (u_k - u_k) + \Lambda_k\,w_k) }$$ So #!latex (-) $$\displaystyle (3.23)\quad \bf{P_{k+1}^-} = \it{E}[ \bf{ \Phi_k (\hat x_k - x_k)^2 \Phi_k^T + \Lambda_k\,w_k^2\,\Lambda_k^T + \Phi_k (\hat x_k - x_k) w_k^2\,\Lambda_k^T + \Lambda_k\,w_k (\hat x_k - x_k)^2 \Phi_k^T } ]$$ In [(3.23)](#EqNum3Dot23) the expectation of the cross terms is zero because (**x**k) contains terms of (**w**k-1) only, and by the assumption that (**w**) is white noise. #!latex (-) $$\displaystyle (3.24)\quad \it{E}[ \bf{ w_k\,w_{k-1} } ] = \bf 0$$ So finally #!latex (-) $$\displaystyle (3.25)\quad \bf{P_{k+1}^-} = \bf{ \Phi_k\,P_k\,\Phi_k^T + \Lambda_k\,Q_k\,\Lambda_k^T }$$ The derivation of the equations in [table(2)](#TblNum2) is now complete □