Fall 2001 16.31 16–17
Interpretations
? Withnoiseinthesystem,themodelisoftheform:
˙x = Ax+Bu+B
w
w, y= Cx+v
– Andtheestimatorisoftheform:
˙
?x = A?x+Bu+L(y??y) , ?y = C?x
? Analysis: inthiscase:
˙
?x =˙x?
˙
?x=[Ax+Bu+B
w
w]?[A?x+Bu+L(y??y)]
= A(x??x)?L(Cx?C?x)+B
w
w?Lv
= A?x?LC?x+B
w
w?Lv
=(A?LC)?x+B
w
w?Lv
? Thisequationoftheestimationerrorexplicitlyshowsthecon?ict
intheestimatordesignprocess. Mustbalancebetween:
– Speedoftheestimatordecayrate,whichisgovernedby λ
i
(A?
LC)
– Impactofthesensingnoise v throughthegain L
? Fast state reconstruction requires rapid decay rate (typically re-
quiresa large L), but thattends tomagnifythee?ect of v onthe
estimationprocess.
– Thee?ectoftheprocessnoiseisalwaysthere,butthechoiceof
L willtendtomitigate/accentuatethee?ectof v on ?x(t).
? Kalman Filterprovidesanoptimalbalancebetweenthetwocon-
?ictingproblemsforagiven“size”oftheprocessandsensingnoises.
Fall 2001 16.31 16–18
? Filter Interpretation: Recallthat
˙
?x =(A?LC)?x+Ly
? Consider a scalar system, and take the Laplace transform of both
sidestoget:
?
X(s)
Y(s)
=
L
sI ?(A?LC)
? Thisisthetransferfunctionfromthe“measurement”tothe“esti-
matedstate”
– Itlookslikealow-pass?lter.
? Clearly, by lowering r, and thus increasing L, we are pushing out
thepole.
– DCgainasymptotesto1/C as L→∞
10
?1
10
0
10
1
10
2
10
3
10
4
10
5
10
6
10
?2
10
?1
10
0
Scalar TF from Y to \hat X for larger L
Freq (rad/sec)
|\hat X / Y|
Increasing L
Fall 2001 16.31 16–19
? Second example: LightlyDampedHarmonicOscillator
bracketleftbigg
˙x
1
˙x
2
bracketrightbigg
=
bracketleftbigg
01
?ω
2
0
0
bracketrightbiggbracketleftbigg
x
1
x
2
bracketrightbigg
+
bracketleftbigg
0
1
bracketrightbigg
w
y = x
1
+v
where R
w
=1andR
v
= r.
? Can sense the position state of the oscillator, but want
to develop an estimator to reconstruct the velocity state.
? Findthelocationoftheoptimalpoles.
G
yw
(s)=
bracketleftbig
10
bracketrightbig
bracketleftbigg
s ?1
ω
2
0
s
bracketrightbigg
?1
bracketleftbigg
0
1
bracketrightbigg
=
1
s
2
+ω
2
0
=
b(s)
a(s)
? Sowemust?ndtheLHProotsof
bracketleftbig
s
2
+ω
2
0
bracketrightbigbracketleftbig
(?s)
2
+ω
2
0
bracketrightbig
+
1
r
=(s
2
+ω
2
0
)
2
+
1
r
=0
?1 ?0.8 ?0.6 ?0.4 ?0.2 0 0.2 0.4 0.6 0.8 1
?1.5
?1
?0.5
0
0.5
1
1.5
Real Axis
Imag Axis
Symmetric root locus
? Notethatas r →0(cleansensor),theestimatorpolestendsto∞
alongthe±45degasymptotes,sothepolesareapproximately
s ≈
?1±j
√
r
? Φ
e
(s)=s
2
+
2
√
r
s+
2
r
=0
Fall 2001 16.31 16–20
? Canusetheseestimatepolelocationsin acker,togetthat
L =
parenleftBigg
bracketleftbigg
01
?ω
2
0
0
bracketrightbigg
2
+
2
√
r
bracketleftbigg
01
?ω
2
0
0
bracketrightbigg
+
2
r
I
parenrightBigg
bracketleftbigg
C
CA
bracketrightbigg
?1
bracketleftbigg
0
1
bracketrightbigg
=
bracketleftBigg
2
r
?ω
2
0
2
√
r
?
2
√
r
ω
2
0
2
r
?ω
2
0
bracketrightBigg
bracketleftbigg bracketleftbig
10
bracketrightbig
bracketleftbig
01
bracketrightbig
bracketrightbigg
?1
bracketleftbigg
0
1
bracketrightbigg
=
bracketleftBigg
2
√
r
2
r
?ω
2
0
bracketrightBigg
? Given L, A,andC,wecandeveloptheestimatortransferfunction
fromthemeasurement y tothe ?x
2
?x
2
y
=
bracketleftbig
01
bracketrightbig
parenleftBigg
sI ?
bracketleftbigg
01
?ω
2
0
0
bracketrightbigg
+
bracketleftBigg
2
√
r
2
r
?ω
2
0
bracketrightBigg
bracketleftbig
10
bracketrightbig
parenrightBigg
?1
bracketleftBigg
2
√
r
2
r
?ω
2
0
bracketrightBigg
=
bracketleftbig
01
bracketrightbig
bracketleftBigg
s+
2
√
r
?1
2
r
s
bracketrightBigg
?1
bracketleftBigg
2
√
r
2
r
?ω
2
0
bracketrightBigg
=
bracketleftbig
01
bracketrightbig
bracketleftBigg
s 1
?2
r
s+
2
√
r
bracketrightBiggbracketleftBigg
2
√
r
2
r
?ω
2
0
bracketrightBigg
1
s
2
+
2
√
r
s+
2
r
=
?2
r
2
√
r
+(s+
2
√
r
)(
2
r
?ω
2
0
)
s
2
+
2
√
r
s+
2
r
≈
s?
√
rω
2
0
s
2
+
2
√
r
s+
2
r
? Filterzeroasymptotesto s =0asr →0andthetwopoles→∞
? Resultingestimatorlookslikea“band-limited”di?erentiator.
– This was expected because we measure position and want to
estimatevelocity.
– Frequency band over which we are willing to perform the dif-
ferentiationdeterminedbythe“relativecleanliness”ofthemea-
surements.
Fall 2001 16.31 16–21
10
?3
10
?2
10
?1
10
0
10
1
10
2
10
3
10
?4
10
?2
10
0
10
2
10
4
Freq (rad/sec)
Mag
Vel sens to Pos state, sen noise r=0.01
10
?3
10
?2
10
?1
10
0
10
1
10
2
10
3
0
50
100
150
200
Freq (rad/sec)
Phase (deg)
10
?3
10
?2
10
?1
10
0
10
1
10
2
10
3
10
?4
10
?2
10
0
10
2
10
4
Freq (rad/sec)
Mag
Vel sens to Pos state, sen noise r=0.0001
10
?3
10
?2
10
?1
10
0
10
1
10
2
10
3
0
50
100
150
200
Freq (rad/sec)
Phase (deg)
10
?3
10
?2
10
?1
10
0
10
1
10
2
10
3
10
?4
10
?2
10
0
10
2
10
4
Freq (rad/sec)
Mag
Vel sens to Pos state, sen noise r=1e?006
10
?3
10
?2
10
?1
10
0
10
1
10
2
10
3
0
50
100
150
200
Freq (rad/sec)
Phase (deg)
Vel sens to Pos state, sen noise r=1e?006
10
?3
10
?2
10
?1
10
0
10
1
10
2
10
3
10
?4
10
?2
10
0
10
2
10
4
Freq (rad/sec)
Mag
Vel sens to Pos state, sen noise r=1e?008
10
?3
10
?2
10
?1
10
0
10
1
10
2
10
3
0
50
100
150
200
Freq (rad/sec)
Phase (deg)