Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Towards
Functional
Safety in
Drive-by-Wire
Vehicles
Towards Functional Safety
in Drive-by-Wire Vehicles
www.TechnicalBooksPDF.com
Peter Johannes Bergmiller
123
www.TechnicalBooksPDF.com
Peter Johannes Bergmiller
Institut für Regelungstechnik
TU Braunschweig
Braunschweig, Bayern
Germany
www.TechnicalBooksPDF.com
v
www.TechnicalBooksPDF.com
vi
www.TechnicalBooksPDF.com
I. I NTRODUCTION
vii
www.TechnicalBooksPDF.com
III. A N OVEL EE A RCHITECTURE FOR D RIVE - BY-W IRE
viii
www.TechnicalBooksPDF.com
V. E VALUATION
ix
www.TechnicalBooksPDF.com
xi
www.TechnicalBooksPDF.com
xiii
www.TechnicalBooksPDF.com
xiv
www.TechnicalBooksPDF.com
www.TechnicalBooksPDF.com
1
719, 535
Increasing
complexity
351 of modern
20 vehicles
New
functionality
vs. safety
www.TechnicalBooksPDF.com
The vehicle
as one
system
www.TechnicalBooksPDF.com
2
Automotive
systems
engineering
Process
models
www.TechnicalBooksPDF.com
The V-Model
The
V-model for
mechatronic
systems
www.TechnicalBooksPDF.com
Requirements Installation,
specification acceptance test
System System
specification validation
Architectural System
design integration and test
Subsystem Subsystem
design integration and test
Module Module
design testing
Coding
Methods
www.TechnicalBooksPDF.com
Customer demands Customer acceptance
Syst
vel
System requirement 1 System acceptance
em le
analysis test
em le
3 System integration
Syst
System design and test
vel
2
IT- IT-integration
requirement
analy
and test
Sub
l
leve
t
analysis &
d tes
Mec nd des
design
sis a
n an s
syst
ic
tem
hanic ign
integ echan
em le
and tegr.
HW- sign
SW- esign
and tegr
bsys
ratio
test
test
& de
&d
s
M
in
analy
in
analy
HW
vel
Su
SW
s. detaile
s.
test nent
test .
SW
p
SW
Boa
layo
draft
com
Deta hanics
st
mec
po
com chanics
Com
el
nt te
rd la gram
ut dia
com
t lev
illing
SW
pone
yout,
pone
implement. HW
onen
Me
of
Manufacture
nt le
p
Prototypes
Com
vel
Standards,
legislation
www.TechnicalBooksPDF.com
1
Method
1
< contains *
Principle
1 *
follows >
Process Heuristics
may contain >
1 *
Tools and
solutions
www.TechnicalBooksPDF.com
Toolchain
A novel
system
architecture
10
www.TechnicalBooksPDF.com
System 3 Toolchain for integration testing Degradation
architecture and hierarchical approach to and 6
Fault
derivation safety evaluation 2 , 4 intervention
Mechanisms
and
methods
Summary of
Contribu-
tions
11
www.TechnicalBooksPDF.com
12
www.TechnicalBooksPDF.com
I Introduction
1 Vehicle Electronics: A Challenge for the Automotive Industry
V Evaluation
9 Functional Safety of MOBILE
13
www.TechnicalBooksPDF.com
www.TechnicalBooksPDF.com
3
Commonly
used tools
www.TechnicalBooksPDF.com
Need for
flexible
vehicles
Goal of this
chapter
RWTH
Aachen
18
www.TechnicalBooksPDF.com
280
Mercedes
Benz
Porsche AG
19
www.TechnicalBooksPDF.com
Nissan
2.5
3 3.3
TU
München
254
Stanford
University
20
www.TechnicalBooksPDF.com
Universität
Paderborn
280 60
2.2
The Chinese
University of
Hong Kong
160 90 DLR
2.05 Kanagawa
Institute of
612 Technology
General
Motors
Melbourne
University
21
www.TechnicalBooksPDF.com
NHTSA
Volvo,
Chalmers
University
Volkswagen
AG
2
4
BMW AG
6
7
AUDI AG,
Vehicle-in-
the-Loop
22
www.TechnicalBooksPDF.com
Construction
and
actuators
Trends in
the design
23
www.TechnicalBooksPDF.com
Conclusion
24
www.TechnicalBooksPDF.com
4
Structure of
the toolchain
15
www.TechnicalBooksPDF.com
→
Electric
drive
26
www.TechnicalBooksPDF.com
100
Four-wheel
steering
40
40
27
www.TechnicalBooksPDF.com
43
130
Electro-
mechanical
braking
1
2.1
Power 300
supply
400
48 12
48
12 →
→
28
www.TechnicalBooksPDF.com
User
interface
Differentiation
of MOBILE
15
29
www.TechnicalBooksPDF.com
Pi-Theorem
1 4.86 15
Restrictions
30
www.TechnicalBooksPDF.com
0.4 1.94 1.75 4.86
0.56 2.72 2.72 4.86
16 2000 2000 125
11 53 44 4.86
0.14 0.68 0.64 4.86
ψ̇ ψ̇ ψ̇
β β β
2
5 24.3 10 4.86
250 151875 607.5
2
· 0.44 1299 2952, 45
230 285
762
Set-up of
MAX
31
www.TechnicalBooksPDF.com
→ →
32
www.TechnicalBooksPDF.com
Code
generation
Evaluation
Monitoring
and
visualization
33
www.TechnicalBooksPDF.com
15
34
www.TechnicalBooksPDF.com
Position Calculation Track Data
cur. coordinates, ref. coordinates,
cur. orientation ref. orientation
Virtual Driver
Speed profile generator Orientation Lateral displacement
controller controller
Speed controller Dynamic weighting
torque steering angle
Longitudinal
Torque
Speed accel- Yaw rate Side slip
vectoring
controller leration controller controller
controller
controller
Dynamic Dynamic
weighting weighting
Torque
distribution
Double track
Drive train model Tire model
model
35
www.TechnicalBooksPDF.com
Double track
vehicle
model
Approach
100
36
www.TechnicalBooksPDF.com
l0...4: Position along the optical lever relative to the vehicle
e0: Yaw angle error
e1...4: Lateral displacement at a given position
Optical lever
e4
Vehicle l1 l2 l3
e0 e e e l4
1 2 3
l0
Reference trajectory
Tuning the
driver
Evaluation
37
www.TechnicalBooksPDF.com
Side slip
Speed
15
38
www.TechnicalBooksPDF.com
www.TechnicalBooksPDF.com
5
Flexibility vs.
safety
Structured
architecture
derivation
1
2
www.TechnicalBooksPDF.com
Derive requirements
1
and constraints
Iteration no
5
finished?
yes
3 5
42
www.TechnicalBooksPDF.com
Vehicle
layer
System layer
(axle control system)
Subsystem layer
(primary axle control system)
Component layer
(acceleration sensor evaluation)
Elementary functions
(calculate average value function)
System
structure
43
www.TechnicalBooksPDF.com
System structure, degree of redundancy 1
Actuator control User interface
ECU design, 2
fault tolerant units
Functional 6
4 Inter-ECU Sensor data 3
redundancy 5
acquisition and
communication Operating
actuator control
system
Power
Power supply
supply 7
Diagnostic algorithms/system
Diagnostic system, fault handling 8
Knowledge
Knowledge base
representation 9
Redundant
local
controllers
44
www.TechnicalBooksPDF.com
Network
centric
approaches
Sensor and
actuator
redundancy
45
www.TechnicalBooksPDF.com
Network
17.6
Operating
systems
Functional
redundancy
Power
supply
100
46
www.TechnicalBooksPDF.com
Online
diagnostics,
degradation
Knowledge
base
Differentiation
of MOBILE
47
www.TechnicalBooksPDF.com
SIRIUS
2001
SPARC
project
Lessons
learned
48
www.TechnicalBooksPDF.com
Control Commands
inputs Vehicle control to actuators
Feedback function Sensor data
Vehicle
Driver
35
49
www.TechnicalBooksPDF.com
Vehicle control function
Emergency run
Control driving function Stability
inputs control Access Commands
Vehicle state control to actuators
acquisition
Diagnostic
function
Prototype Sensor
Feedback data
driving function
Functional
architecture
50
www.TechnicalBooksPDF.com
Front axle User Rear axle
control system interface control system
Data bus
Hardware
structure
Merging
views
51
www.TechnicalBooksPDF.com
User Front axle Rear axle Power Stability Inertial
interface control control supply control meas. sys.
Emergency run driving function Stability Vehicle
control state
Prototype driving function
acquis.
Access control
Diagnostic function
Driving
functions
State
estimation
Stability
control
Diagnostic
system
52
www.TechnicalBooksPDF.com
Vehicle control function
Emergency run
Emergency run driving function driving function
Control
Local data Data Actuator
inputs
acquisition processing control
Switch Commands
to actuators
Access control Prototype
Vehicle state acquisition Stability control
driving function
Inertial Sensor State Reference Control Safety
algorithm guard Actuator
measurm. data fusion estimation generation
control
Safety
monitoring
Prototype driving function
www.TechnicalBooksPDF.com
Feedback Feedback Diagnostic function
control
Global data Diagnostic Hardware
acquisition fusion monitoring
Sensor
data
53
Access
control
Power
supply
Stability
controller
54
www.TechnicalBooksPDF.com
Fault
tolerant
units
Merging
views
55
www.TechnicalBooksPDF.com
56
Primary front axle controller Primary rear axle controller Primary user interface controller Primary power supply controller
Local data Data Actuator Local data Data Actuator Local data Data Actuator
acquisition processing control acquisition processing control acquisition processing control
Feedback
control
Secondary front axle controller Secondary rear axle controller Secondary user interface controller Secondary power supply controller
Local data Data Actuator Local data Data Actuator Local data Data Actuator
acquisition processing control acquisition processing control acquisition processing control
Diagnostic Hardware Diagnostic Hardware Diagnostic Hardware Global data Diagnostic Hardware
fusion monitoring fusion monitoring fusion monitoring acquisition fusion monitoring
Safety Safety Safety Safety Safety Safety Safety Safety Safety Safety Safety Safety
reference monitoring guard reference monitoring guard reference monitoring guard reference monitoring guard
www.TechnicalBooksPDF.com
Stability controller Inertial measurement controller
Vehicle state acquisition
Reference Control Inertial Sensor State
generation algorithm measurm. data fusion estimation
Diagnostic function
Safety
guard Prototype driving function
57
www.TechnicalBooksPDF.com
Application
interface
58
www.TechnicalBooksPDF.com
59
www.TechnicalBooksPDF.com
Assumptions
and future
challenges
Summary
60
www.TechnicalBooksPDF.com
6
www.TechnicalBooksPDF.com
Functional
redundancy
62
www.TechnicalBooksPDF.com
Unit 1
Emergency-off function
Propulsion sub-function
Driver
Actuators
inputs Steering sub-function
Braking sub-function
63
www.TechnicalBooksPDF.com
system
of the
Perception
64
Functional system architecture
and hardware constraints
Drive
... ...
Layers for
functionality
(vehicle level)
Elementary functions
hierarchical abstraction
www.TechnicalBooksPDF.com
Goal orientation
Hardware orientation
Requirements
Failure probabilities
Result
Vehicle layer
layer
System layer
Detailed
(axle control system)
in this
Subsystem layer work
(primary axle control system)
Component layer
(acceleration sensor evaluation) Assumed
Elementary functions inputs
(calculate average value function)
Hierachical
layering in
research
65
www.TechnicalBooksPDF.com
Input: safety requirements
Architecture
analysis 1 Define hierarchical layers
Component
analysis Identify generalized
Adapt if needed
Iteration no
6
finished?
yes
Top-level
functional Derive cut sets for
7
safety relevant failure scenarios
evaluation
66
www.TechnicalBooksPDF.com
Virtual systems and
interfaces
Existing architecture
and interfaces
Examples
Critique
67
www.TechnicalBooksPDF.com
3
Definition
process
68
www.TechnicalBooksPDF.com
In Out
C1 C2 C3 C4
C5
Related
work
69
www.TechnicalBooksPDF.com
70
www.TechnicalBooksPDF.com
1
p21 p12
p31
p13 p11
2 p22
p33
p23 p32
3
pij i
j
Xt si
X i
pij
71
www.TechnicalBooksPDF.com
6
72
www.TechnicalBooksPDF.com
Failure states 1 to 11 Abstraction
(step a) (step b)
Generalized failure states
(step c)
73
www.TechnicalBooksPDF.com
6
74
www.TechnicalBooksPDF.com
Hierarchical layer
(1...L)
Cut set counter for
each layer (1...I) Single- or dual-point fault based cut set
l
i s d
Cl l C
1
12 C12 s
Example
Cil{s/d} = Cjg{s/d} ∀ i = j ∧ l = g,
Cils ∩ Cjgd =∅ ∀ i, j, l, g.
l g i j
Result:
cut sets
75
www.TechnicalBooksPDF.com
Result layer / layer 1
Faulty part of the system
Layer 2
Layer 3
External
Layer 4
λli
i l
TM
λ1 λ2
p
1
p = 2 · λ1 · λ2 · (TM · TSS − · T 2 ),
2 SS
76
www.TechnicalBooksPDF.com
Both systems
operable
Failure of one
system
1-c c
TSS
f (Δt) = 1 − e−λΔt
f (Δt) = λΔt λ
Δt
1
P (Cild ) = 2 · λli,1 · λli,2 · (TM · TSS − · T 2 ).
2 SS
λli,1 λli,2
i
l
77
www.TechnicalBooksPDF.com
l
I
P (C l ) =(+1) · P (Cil{s/d} )
i=0
i=I−1 j=I
(−1) · P (Cil{s/d} ∩ Cjl{s/d} ) } (λ3 ) ≈ 0
i=0 j=i+1
j=I−1
i=I−2 k=I
(+1) · P (Cil{s/d} ∩ Cjl{s/d} ∩ Ckl {s/d} ) } (λ4 ) ≈ 0
i=0 j=i+1 k=j+1
...
(−1) I+1
· P (C1l {s/d} ∩ C2l {s/d} ∩ ... ∩ CIl{s/d} ) } (λI+1 ) ≈ 0.
I j k
i (·)
λ
λ
i j
λli,1 λlj,1
λ2
8 F
P (C l )
l=L
F = P (C l ).
l=1
78
www.TechnicalBooksPDF.com
DC
λdd
DC = .
λdd + λdu
λdd λdu
Calculation
process
λdu DC
λdu
λdd
79
www.TechnicalBooksPDF.com
Fields of
application
Contributions
80
www.TechnicalBooksPDF.com
81
www.TechnicalBooksPDF.com
Outlook
82
www.TechnicalBooksPDF.com
www.TechnicalBooksPDF.com
7
“Zebra
tactics” for
MOBILE
www.TechnicalBooksPDF.com
86
www.TechnicalBooksPDF.com
Requirements
Solution
strategy
→ →
→
→
→
87
www.TechnicalBooksPDF.com
→
→
→
Core
contributions
(1) System
structures
88
www.TechnicalBooksPDF.com
(2) Deriving
symptoms
1970
(3)
Determining
actions
89
www.TechnicalBooksPDF.com
Cause-
symptom
correlation
Fuzzy Logic
1960
Bayesian
Networks
90
www.TechnicalBooksPDF.com
Neural
Networks
1943 1980
91
www.TechnicalBooksPDF.com
(4) Relating
PFDH
→
→
<4
→
→ →
S Ns
92
www.TechnicalBooksPDF.com
Controller
11 1 11 2 3
21 21 Symptom 1
Input
... Signal
...
signals
N1 N1
group p1
1
12 12
22 22 Signal Symptom 2 Action
Input Action
Monitoring
...
...
signals group p2 derivation
N2 unit N2 unit
2
1S 1S
2S 2S Symptom S
Input Signal
...
...
signals
NS NS group pS
S
93
www.TechnicalBooksPDF.com
ps
n=Ns
n=1 (hs,n · psn )
ps = max n=Ns
, Es , s ∈ {1..S}, n ∈ {1..Ns }, hs,n ∈ R+ ,
n=1 (hs,n )
hsn n s
Ns psn n Es
(A) - Error
perpetrator
s ps
Ps ( s ∩ s )=
ps · Ps ( s | s ).
Ps ( s s )
0.5
Ps ( s ∩
s ) s
s
(B) - Signal
group error
probability
e S
vectors
(C) - Cause
identification
94
R
S cer,s
1 0
cr
c = (c1 , .., cR )T
s=S
s=1 cer,s · es
cr = s=S
.
s=1 cer,s
c
(D) - Action
derivation
G R
acg,r g r
ag
r=R
ag = (acg,r · cr + (1 − cr )).
r=1
(1 − cr )
g (acg,r · cr )
a = (a1 , .., aG )T
(E) - Action
weighting
95
No fault Weighted costs
Vehicle operable
after application of
an action
ug
wg
wg
r=R
r=1 acg,r · (1 − cr )
wg = r=R ,
maxg ( r=1 acg,r · (1 − cr ))
ug
r=R
ug = wg · (1 − cr ).
r=1
ug u a u
v = a − u g
vg
(F) - Time
dependency
g
g
96
c (th )
th
c (th ) c
c(th ) c h
c
c (th )
c(th )− = 0
c (th )
97
User inputs
(gas + steering),
sensor information
FlexRay A FlexRay B
Output
Input Secondary units
units node Input
units
Actuator
Input Primary Output
Sensor data commands
units node units
98
n
p sn = 0 ps n = 1
4
2 10
Internal
health
monitoring
99
Prop
Limit
Prop
Limit
Del
Del
DelayMon X O O ... No action 0 0 0 ...
PropMon O X O ... SW reset 0.5 0.5 0.5 ...
FrequMon O O O ... HW reset 0.8 0.8 0.8 ...
...
...
...
...
...
...
CE of SN AC
IntSysTim
CANAFail
DevSW
Prop
Limit
Del ...
DevSW X X X No action 0 0 0 ...
IntSysTim X X O ... SW reset 0.2 0.8 0.1 ...
CANAFail O O O ... Power off 1 1 1 ...
...
...
...
CE of PN ACof PN
2 10
Matrices
100
250
101
10% 33%
t1
t2
t3
t4
pdel /pprop
p 1 p2 p3 0 0.10
102
1 1
Probability values Probability values
Action probability
Action probability
0 OR 0.10 0 OR 0.30
0.5 0.5
0 0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Cycle Cycle
1 1
Probability values Probability values
Action probability
Action probability
0 OR 0.60 0 OR 0.90
0.5 0.5
0 0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Cycle Cycle
SN&PN NoAction
SN SW−Reset
SN HW−Reset
SN PowerOff
PN SW−Reset
PN PowerOff
p1 p2 p3
(0, 0, 0)
Discussion
of simulation
results
103
4
104
1 1
Cycle #1 Cycle #2
Action probability
Action probability
0.5 0.5
0 0
0 0.5 1 0 0.5 1
Probability input Probability input
SN & PN NoAction
SN SW−Reset
SN HW−Reset
SN PowerOff
PN SW−Reset
PN PowerOff
PFDH and
Bayesian
Networks
147
105
SG
SG1 SG2 ... SGN
1
SN
CoF CoF CoF CoF
... PN
...
SN1 SNR PN1 PNT
ASr,g / APt,z : efficiency of action g/z to address r/t -th cause on the
secondary/primary node
CSr,g / CPt,z : costs generated by action g/z on cause r/t on the
secondary/primary node
CoF SN/PNr,t : r/t-th cause of failure on the monitoring/development node
ASG SN/PNr,t: signal group failure rate associated to the
secondary/primary node
SGn: probability of failure detected by signal group n
25%
97%
3%
106
107
108
Increase
benefits,
reduce costs
109
Model-
following
controller
Controller
onboard
15
MAX
100%
110
5
Measured acceleration
Acceleration (m/s²), speed (m/s)
4 Reference acceleration
Measured speed
Reference speed
3
−1
0 5 10 15 20 25 30 35 40 45 50
Time (s)
Failure com-
pensation
111
2
Yaw rate (rad/s), side slip angle (rad)
−1
Measured yaw rate
Reference yaw rate
−2
Measured side slip angle
Reference side slip angle
−3
0 5 10 15 20 25 30 35 40 45 50
Time (s)
Conclusion
112
113
8
Strategic
mechanisms
for MOBILE
116
Requirements
Solution
strategy
117
Critical wear level 1 without optimization
Proportional wear
2 with optimization
(wear balancing)
3 with optimization
(overall wear
reduction)
118
Need for
coordination
Modeling
components
Impact
factors on
tire wear
119
Δm = f1 · W f2 .
f1 f2
Learning
from mea-
surements
W = FL · (Ω · R · cos(α) − ω · r) + FT · Ω · R sin(α).
FL FT
Ω·R
α ω·r
Modeling
camber
120
Pressure
distribution Trapezoid due to camber
Non-
isotropic
shear forces
Optimizing
tire load
121
μ
→ →
Integrating
the optimizer
122
Driver
Constraints
commands
Low execution
frequency
Optimizer
High execution
frequency
Drive Distribution Vehicle
controller unit (actuators)
Execution
frequency
The driver
Functional
safety
p1 pQ
123
Optimizer results
Basic constraint
Commands to actuators
Access control
Fall-back
p1
p2
...
pQ
min z = f (x),
124
x = (x1 , x2 , ..., xG ).
min z
f x
gi hj I J xu
xl x
f1 fN N
Evaluation
Evolutionary
algorithms
125
NSGA-II
x
126
Gradient
based
algorithms
min z
fn (x) n
n
N
min z = (|fn (x) − n| · n ).
n=1
N 1/p
min z = lim (fn (x))p .
p→∞
n=1
SQP
127
Tmax
p1 pQ
xg
128
nM (t) t
g=G
Mset (t) − g=1 Mg (t)
nM (t) = ,
Mset (t)
Mg (t) g G
Mset (t)
129
F Ni i = 1..4
sl,i
sr,i
μ {r,l},i
= min(C · s{r,l},i , μ ).
μ μ {r,l},i
Pi
vl,i vr,i
130
i Pr,i Pl,i
Pr,i = μ r,i
· FNi · vr,i ,
Pl,i = μ l,i
· FNi · vl,i ,
Pi = Pr,i + Pl,i .
Pi Wi
Wi (t)
Wi (t) − Wmin (t)
nWi (t) = 0.1 + 0.9 · .
Wmax (t) − Wmin (t)
Wmax (t) Wmin (t)
t
t
T (t)
T (t + Δt) M Δt
T (t + Δt) − Tlow
nT (t) = .
Tmax − Tlow
Tlow Tmax
13 − 15
131
2
1.8
Weight (−)
1.6
1.4
1.2
1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Input (−)
nB
1
(η) = 1 + η .
e−(K1 · 100 −K2 ) + K3
K1 K2 K3
η
K1 = 10 K2 = 8
K3 = 0.86
132
1
1 + K3
1 2
0 1
=1
25
nM = 50 − 25 · max({ }).
133
60
50
Y coordinate (m)
40
30
20
10 Start at (0,0) in
counterclockwise
direction
0
−70 −60 −50 −40 −30 −20 −10 0 10 20
X coordinate (m)
1.1
20
0.15 0.1
1.6
134
6
5
Speed (m/s)
0
0 50 100 150 200 250 300 350
Track (m)
80
100 36
0.15
Measure-
ments with
MAX
∀i w(i)
1−5
σ̄
σ̄
Ŵ
135
Steering angle
0.35
0.3
Steering angle (rad)
0.25
0.2
Left (Ackermann)
0.15 Right (Ackermann)
0.1 Left (optimizer)
Right (optimizer)
0.05
0
0 5 10 15 20 25 30 35
Track (m)
Wear of tires (front left, front right, mean value of rear axle)
350
Tire wear, energy estimate (J)
300
250
200
Tire wear with optimizer
150 Tire wear with Ackermann steering
100
0 5 10 15 20 25 30 35
Track (m)
100
56
2
4 3.5
136
∀i w(i) σ̄ σ̄ Ŵ
∀i w(i) σ̄ σ̄
Ŵ
350
6.5
30
25
70
137
Speed
6
Speed (m/s)
4
Reference
Without optimizer
2
With optimizer
0
0 50 100 150 200 250 300 350 400
Track (m)
Torque front and rear axle
100
Torque without optimizer front/rear
Torque command (%)
40
20
0
0 50 100 150 200 250 300 350 400
Track (m)
25
70
260
138
Temperatures
100
80
Temperature (°C)
60
Without optimizer front
40 Without optimizer rear
With optimizer front
20 With optimizer rear
Critical temperature
0
0 50 100 150 200 250 300 350 400
Track (m)
States of Charge
100
Without optimizer front
State of Charge (%)
40
20
0
0 50 100 150 200 250 300 350 400
Track (m)
v̄
¯
|d|
T̄
Total wear
139
Temperatures
100
Without optimizer
80 With optimizer front
Temperature (°C)
40
20
0
0 50 100 150 200 250 300 350
Track (m)
Tire wear front left, front right, and rear axle (Ackermann)
3000
Tire wear front left, front right, and rear axle (optimizer)
2500
2000
1500
1000
500
0
0 50 100 150 200 250 300 350
Track (m)
96
105
15
9 131
140
¯
∀i w(i) σ̄ σ̄ Ŵ v̄ |d| T̄
+10
+10
∀i w(i) σ̄ σ̄
Ŵ v̄
¯
|d| T̄
93
95 Driver inputs
Standard
deviation of
wear
141
Peak wear 10
and temper-
atures
6 8
Failure
scenarios
Unequal
wear 1.1 10
Ŵ
10
Summary
for complex
track
142
400
350
Standard deviation relative to
non−optimized driving (%)
300
250
200
150
100
Measurements
50 Vehicle with Ackermann steering (reference)
0
0 0.5 1 1.5 2 2.5
Period between optimizer executions (s)
0.2
0.4
143
Optimization
Strategy
Limitations
of the
system
Possible
further de-
velopments
144
Self-
concept
145
Self-esteem
10
146
Wisdom Further abstraction, understanding principles
Self-repre-
sentation
Further
definitions
Data
147
Information
Knowledge
148
Abilities and
skills
149
Behav-
ioral skills
Action skills Skills
Bas kil s
kil s
Act kil s
iora av-
s
ic s
ls
Beh
ion
Basic skills
e
tis Abilities
er
s
ie
xp
ilit
fe
ilit tive
ab
lo
M s
ie
ab gni
or
ve
ot
Le
o
C
Transfer to a
vehicle
Hierarchical
abstraction
150
Level of expertise Self esteem
of the vehicle
Self concept Vehicle Driver
Cognitive abilities
Behav- Experts
Navigation
ioral skills
Action skills Guidance
Relating
results
151
152
Summary of
Contribu-
tions
Universität
der Bundes-
wehr
München
153
Project
“Stadtpilot”
Project
SPARC
154
BMW-Group
Applied
Research
Laboratory
155
Autonomous
robots
Motor and
perceptual
schemas
Cognitive
automobiles
156
Summary
157
Requirements
Solution
strategy
Information
base
158
→
→
→
Knowledge
→ base
159
... Onboard Offboard
Device 1 Device N
→ →
Core
contributions
160
Information
base
161
Configuring
the
information
base
Knowledge
base
162
al
ior
h av
Be s Fuzzy Fuzzy
ill
Sk
n
tio
Ac s Fuzzy Fuzzy Fuzzy Fuzzy
k ill
S
s ic
Ba s Fuzzy Fuzzy Fuzzy Fuzzy Fuzzy
ill
Sk d
he
ric on
En mati
o r
f on
in ati
f o rm
In
se
ba
163
A fuzzy way
to determine
skill levels
Evaluation
164
Sharp inputs Fuzzy Logic in a skill node
(measurements, Sharp output
other skills etc.) Fuzzify Inference Defuzzify of skill level
165
Behavioral skills
City driving
Demo Highway ...
driving driving
Action skills
Stop&go
Lane
following
Maneuver-
ing
Lane
change
High-speed ...
driving
Basic skills
Set driving
Standstill Decelerate Accelerate
direction
Provide Provide
Parking Rotate
power energy
Implementa-
tion
166
Technical
facts
4 20
0.5 1
Concept:
information
base
Concept:
knowledge
base
167
Outlook
Tooling
168
Transfer to
series
vehicles
Conclusion
169
9
Safe state
Note:
Remark for
series
vehicles
174
1
6 6
7 10
5 6
13.9 50
6 10
0.6
0.35 0.09
20
175
≤ 50
176
30
30
30
160 44
4
30
177
2 9
31
702
60
“Subsystem
layer”
100
178
Lower layers
Link
between
layers
Summary
1140
179
5,5E-08
5E-08
Faillure rate (1/h)
4,5E-08
Failure rate on
vehicle level
4E-08
3,5E-08
3E-08
114
171
228
285
342
399
456
513
570
627
684
741
798
855
912
969
1026
1083
1140
0
57
0,008
Failure rate on
0,006 system level
Fa
0 004
0,004
Failure rate on
0,002 subsystem level
0
171
228
285
342
399
456
513
570
627
684
741
798
855
912
969
1026
1083
0
57
114
1140
180
10
System
architecture
Tactical
mechanisms
182
A hierarchical approach to
functional safety evaluation
Self-representation
Strategic measures
Integrated vehicle
control to exploit
Deriving a system
architecture top-down
functional redundanies
Handling Algorithm
Functional requirements in terms of application and safety
15 Validation
and
verification
Summary
183
Erratum to: Towards Functional Safety
in Drive-by-Wire Vehicles
Erratum to:
P.J. Bergmiller, Towards Functional Safety
in Drive-by-Wire Vehicles,
DOI 10.1007/978-3-319-17485-3
Economic
Commission
for Europe
407/2011
ISO, IEC,
ITU, EN and
DIN
Legislation
in Germany
823
186
25
187
188
Π
Π
Π
Ψ̇
mv β̇ + (mv 2 + cαf lf − cαr lr ) + (cαf + cαr )β = cαf δf + cαr δr
v
Ψ̇
ΘΨ̈ + (cαf lf2 + cαr lr2 ) + (cαf lf − cαr lr )β = cαf lf δf − cαr lr δr
v
13
10 Π
m
m kg v s
Θ kgm2 δf /r rad
rad
β rad β̇ s
Ψ̇ rad
s
Ψ̈ rad
s2
lf /r m
kgm
cf /r s2 rad
189
Π
β̇lf
Π1 =
v
Ψ̇lf
Π2 =
v
Ψ̈lf2
Π3 =
v2
Π4 = β
Π5 = δf
Π6 = δr
Θ
Π7 =
lf2 m
cf l
Π8 =
v2m
cr l
Π9 =
v2m
lr
Π10 =
lf
Π
190
P
s| r)
P( r| x ∩ ...∩ y ))
191
192
193
p
N ∈ N
n pn
1 1 1
p = · p1 + · p2 + · · · + · pn .
N N N
N =1
1
p = p1
1
194
N =2
w1
w1 +w2
w2
w1 +w2
w 1 , w2 ∈ N
p1 p2 p
w1
w1 +w2
w2
w1 +w2
p
w1 w2
p = p1 · p2 + · p1 · (1 − p2 ) + · (1 − p1 ) · p2
w 1 + w2 w 1 + w2
w1 w2
= · p1 + · p2 .
w 1 + w2 w 1 + w2
w1 = 1, w2 = 1 w1 + w 2 = N = 2
1 1
p = · p1 + · p2
2 2
n→n+1
w1 w2prelimn
p = · p1 + · p2prelimn ,
w1 + w2prelimn w1 + w2prelimn
p2prelimn
w2 w3
p2prelimn = · p2 + · p3 ,
w 2 + w3 w 2 + w3
w2 + w3 = w2prelimn
N
w1 w2 + w3 w2 w3
p = · p1 + ·( · p2 + · p3 )
w 1 + w2 + w3 w 1 + w2 + w3 w 2 + w3 w2 + w3
w1 w2 w3
= · p1 + · p2 + · p3 .
w 1 + w2 + w3 w 1 + w2 + w3 w 1 + w 2 + w3
195
= 1
n n+1
N
1 1 1
p = · p1 + · p2 + · · · + · pN
N N N
N
p1 p2 ··· pN p
w1new +w2new +···+wNnew
{0, 1} {0, 1} · · · {0, 1} w1 +w2 +···+wN
wnnew 0 pn wn
i=N
w n ∈ R+ i=1 wi =
N p
w1 w2 wN
p = · p1 + · p2 + · · · + · pN .
N N N
N
2N
N
2N 2N − 1
N N −1
196
N ∈N
N + (N − 1) > (N − 1) ∗ 2N + (2N − 1)
N ∈N
30
Operations per node
(N−1)*2N+(2N−1)
20
10
0 N+(N−1)
0 1 2 3
Number of parents (N)
40
197
1.5 2.7
1.6 100
0 100 10
100 40
Fuzzify
Inference
0 1
198
too very too
low low low medium high high
1
Membership
0.5
0
Minimal Maximal
value value
Rule 1
0.5
Rule 2
Merged
results
0
25 50 75 100
Skill level
Defuzzify
0 25 50 75 100 75
199
High-voltage
components
200
Electronics
12 48
Vehicle
frame
Integration
201
Summary
202
36 8
Future
extensions
2.4
203
B