how traffic flow simulation can teach automated · pdf filehow traffic flow simulation can...
TRANSCRIPT
![Page 1: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/1.jpg)
1
1S C I E N C E � P A S S I O N � T E C H N O L O G Y
� www.tugraz.at
ptv Scientific Advisory BoardKarlsruhe Jul 13, 2017
How traffic flow simulation can teach automated driving
Prof. Martin FELLENDORFGraz University of Technology
Inst. f. Transport Planning and Highway Engineeringwww.isv.tugraz.at
![Page 2: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/2.jpg)
2
2
The world is changing ....
2005 2013
Michael Sohn /APLuca Bruno /AP
![Page 3: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/3.jpg)
3
3
Mobility Buzz Words
Demographic change & urbanization Electrification ���� Decarbonization
Sharing mobility ���� MaaS Assisted ���� Autonomous ���� Connected
![Page 4: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/4.jpg)
4
4
Car commercials
Land-Rover: https://inspiredpixel.files.wordpress.com/2011/04/land-rover-photography-sport-1.jpg
Mercedes: https://theadsgarage.com/tag/commercial/page/2/
BMW: http://dirs-n-dops.de/manuel-wenger.html
Ford: https://www.ford.com/finance/commercial-financing
Porsche: https://www.autoevolution.com/news/porsche-macan-commercial-believer-video-79000.html
![Page 5: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/5.jpg)
5
5
.. and reality
![Page 6: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/6.jpg)
6
6
Expectations of Connected & Automated Vehicles (CAV´s)
Efficieny• capacity increase on
motorways, shorter headways
• higher saturation flow at signals
• Friedrich, B; Wagner, P. ;...
Safety
• lower risk
• 90% accid. human related
Environment
• less vehicles (MaaS)
• efficient driving
![Page 7: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/7.jpg)
7
7
Capacity
Sp
ee
d [km
/h]
Traffic Volume [veh/h]
Conventional cars
• human behavior• unequal headways• heterogeneous
traffic flow
Lveh Lveh Lveh Lvehh1 h2 h3
Automated cars
• equal headways• shorter latency (V2V)• homogenous traffic flow
• higher density?• higher speed?
• higher capacity?• increased stability?• higher throughput?
Lveh Lveh Lveh Lveh Lvehh h h h
Source: Neuhold, Fellendorf: MobilTUM, 2015
![Page 8: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/8.jpg)
8
8
capacity: simple calculation
Vehicles need space: � � �� � ℓ��
� � velocity, � headway, ℓ�� vehicle length
� Thus: ���
��
���
��ℓ����: � vehicles per kilometer per lane
� Flow � � ��, with � ����
��ℓ���
Some values:
� ℓ�� � 6�, � � 1.8�, � � 22m/s (! 80km/h)
� � � 20veh/km and # � max �& � 1740veh/h
and automated
� � � 0.5� → � � 87veh/km and # � 6886 veh/h
� � � 0 → � � 133veh/km and # � 13200 veh/h ℓ+,-
./
lead
lag
![Page 9: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/9.jpg)
9
9
BUT simplification does not meet reality
single lane
� car following model
� longitudinal movement in Vissim
multilane infrastructure
� motorways (eg. merge/diverge)
� urban arterials (eg. intersections)
� longitudinal and lateral movement
Adaptive Cruise Control (ACC)
� car following
Advanced driver-assistance Systems (ADAS)
� car following
� including lane change
© www.automotive-technology.de
![Page 10: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/10.jpg)
10
10
Capacity studies on ADAS and CAV´s
among others:
FAT 296
� TU Munich and KIT: Capacity impact of partial and highly
automated driving,
� Study for the German Automobile Association (2014-2016)
� Vissim with adjusted longitudinal and lateral parameter values
https://www.vda.de/de/services/Publ
ikationen/fat-schriftenreihe-296.html
Motamedidehkordi, N.; Margreiter, M., Benz, Th: Effects of Connected Highly Automated Vehicles on the Propagation of
Congested Patterns on Freeways, TRB 16-1802
![Page 11: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/11.jpg)
11
11
CAV´s modelled by parameter changes
^FAT 296, 2017, S 19
![Page 12: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/12.jpg)
12
12
Capacity studies on ADAS and CAV´s
among others:
FAT 296
� TU Munich and KIT: Capacity impact of partial and highly
automated driving,
� Study for the German Automobile Association (2014-2016)
� Vissim with adjusted longitudinal and lateral parameter values
https://www.vda.de/de/services/Publ
ikationen/fat-schriftenreihe-296.html
Haberl, Neuhold, Fellendorf, Rudieger, Kerschbauer, Eichberger, Rogic: Simulation assisted impact analyses of automated
driving on motorways for different Levels of automation and penetration, mobil.TUM 2017, to be published
Motamedidehkordi, N.; Margreiter, M., Benz, Th: Effects of Connected Highly Automated Vehicles on the Propagation of
Congested Patterns on Freeways, TRB 16-1802
VEGAS
� TU Graz and Virtual Vehicle (2016-2017); same topic
� supported by bmvit and Asfinag
� Vissim with driver.dll and driving simulator
![Page 13: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/13.jpg)
13
13
VEGAS
Objective:
� impact of automated driving (AD) on availability of the Austrian motorway network
Methodology
MODEL TEST DETERMINE ANALYSE
Haberl, Rudieger etal. mobil.TUM 2017, to be published
![Page 14: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/14.jpg)
14
14
VEGAS: general approach8 Typical Link Segments
Automated Driving Functions *via MATLAB/Simulink
Changes of Capacity due Automated Drivingvia Microscopic Simulations
Projection of Changed Capacity to
the Network ∆C
∆C
* Driving Simulator Study
∆C
~83% ~83%
Haberl, Rudieger etal. mobil.TUM 2017, to be published
![Page 15: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/15.jpg)
15
15
automatic driving (AD) in VISSIM: driver.dll
Driver.DLL functions
� Init DriverDLL
� Set Data
• VISSIM provides
- EGO-vehicle (pos, v, type, ..)
- neighboring vehicles (pos, v, type, ..)
- network topology (static)
- infrastructure control (dynamic)
� Get Data
• VISSIM receives EGO vehicle
- target acceleration
- Lane data (change active, target lane, angle)
� Kill DriverDLLRun simulation in VISSIM with
various DLL´s
Run simulation in VISSIM with
various DLL´s
Modeling and Testing of AD
systems in MATLAB/Simulink
Convert Simulink code to C++
Convert Simulink code to C++
C++DLLC++DLL
Connect DLL with VISSIM
AD-functional specificationAD-functional specification
Haberl, Rudieger etal. mobil.TUM 2017, to be published
![Page 16: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/16.jpg)
16
16
AD in driver.DLL and state machine for lateral movement
Haberl, Rudieger etal. mobil.TUM 2017, to be published
![Page 17: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/17.jpg)
17
17
Automatic Driving Function as driver.dll
Traffic Jam Assist – Level 2
� at congestion up to 30 km/h with advanced ACC + Stop & Go
� target headway as parameter (0,5s; 0.9s; 1,1s; 1.8 s)
Traffic Jam Chauffeur – Level 3
� at dense traffic up to 60 km/h with advanced ACC + Stop & Go
� target headway as parameter (0,5s; 0.9s; 1,1s; 1.8 s)
Haberl, Rudieger etal. mobil.TUM 2017, to be published
![Page 18: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/18.jpg)
18
18
Automatic Driving Function as driver.dll (cont´d)
Highway Pilot – Level 4
� automated driving on motorways and multilane highways up to 130 km/h
� no ad-hoc platooning as no industry-standard
Truck Platooning – Level 3
� platooning at rightmost lane
� no special handling of trucks leaving platoon
Haberl, Rudieger etal. mobil.TUM 2017, to be published
![Page 19: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/19.jpg)
19
19
Automated Driving Functions (ADF) for SAE2, SAE3 and SAE4
� Model specific ADF for microscopic traffic flow simulation (VISSIM)
Modelled ADF Longitudinal control Lateral control V2V
Traffic Jam Assist – SAE Level 2 VISSIM + Driver-DLL <30km/hVISSIM + Driver-DLL <30km/h (no lane-change)
No
Traffic Jam Chauffeur – SAE Level 3 VISSIM + Driver-DLL <60km/hVISSIM + Driver-DLL <60km/h (no lane-change)
No
Highway Chauffeur – SAE Level 3Driver-DLL: PC 130km/h, HGV 90km/h
Driver-DLL No
Highway Pilot – SAE Level 4 Driver-DLL: PC 130km/h, HGV 90km/h
Driver-DLL Yes
C-ACC Platooning of HGV– SAE Level 1 Driver-DLL VISSIM Driver Yes
Truck Platooning – SAE Level 3 Driver-DLL Driver-DLL Yes
Haberl, Rudieger etal. mobil.TUM 2017, to be published
![Page 20: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/20.jpg)
20
20
Testing typical motorway network elements
� Guideline specific elements (RVS 03.05.13) for merge (M), exit (E) and weaving (W)
2+1RM 3+1RM 2-1RE 3-1RE 2-lane 3-lane 2+1VV 3-1 sub 2+1 add
![Page 21: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/21.jpg)
21
21
Driving Simulator Study
Objective:
� drivers’ acceptance of AV
� Calibrate Vissim parameters
Test site
� 3-lane motorway section of ~12 km with 5 entries
� Vehicle demand 60% of guideline capacity, 23% of vehicles exit/enter at junction
� 10% heavy good vehicles (HGV)
3 Scenarios per person
� Simulator vehicle runs as SAE 0
� SAE 0
� SAE 4: short headway 0.9 sec
� SAE 4: long headway 1.8 sec
Off-ramp in VISSIM
A1, B1, C1
On-ramp in VISSIM
A2, B2, C2
Weaving section in VISSIM
D, E
Haberl, Rudieger etal. mobil.TUM 2017, to be published
![Page 22: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/22.jpg)
22
22
Driving Simulator at TU Graz, Institute of Automotive Engineering
PTWinSimPodium Technology
• Real-time communication of each single module
• Manage start & end of simulation
Traffic SimulationInterface
• Data preparation for application in
the Driving Simulator
• Interpolation of VISSIM 20 Hz to DS 200 Hz
PTV VISSIM DS Interface
• Operate several external vehicles
• Provide positions of simulation
vehicles for DS
© Eichberger et al. (2017)
VISUAL CUEING
ACOUSTIC CUEING
HAPTIC CUEING
MOTION
CUEING
REAL-TIME SIMULATION FRAMEWORK
PHYSIOLOGICALMEASUREMENTS
CUEING(HARDWARE)
CUEING(HARDWARE)
EGO VEHICLE
DYNAMICS
ADAS SENSOR MODELS
ADAS DRIVING
FUNCTIONS
TRAFFIC SIMULATION
SIMULATION (SOFTWARE)SIMULATION (SOFTWARE)
![Page 23: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/23.jpg)
23
23
Test Drivers’ Acceptance of Automated Vehicles
some flaws with Driver.DLL – vehicles
� exessive lane changes
� runs over shoulder
![Page 24: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/24.jpg)
24
24
Driving Simulator – different scenarios
![Page 25: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/25.jpg)
25
25
5 features tested in simulation scenarios
1. feature ( n=8 )Link Segments
2. feature ( n=6 )Penetration Rates
3. feature ( n=3 )Speed
4. feature ( n=3 )HGV Ratios
5. feature ( n=3 )Vehicle Input Ratios
10%10%
15%15%5%5%
Vehicle input ratio Mainline Ramp
G1-2 64% 36%
G3-4 80% 20%
G5-6 91% 9%
© G
eis
tefe
lde
t al. (2
01
5)
Haberl, Neuhold, etal. mobil.TUM 2017, to be published
![Page 26: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/26.jpg)
26
26
Results of Typical Motorway Segments – 2+1 Lanes: SAE0
5000m 300m 1200m
DCP 1 DCP 2 DCP 3 DCP 4
DCP 1 DCP 2 DCP 3 DCP 4
DCP = Data Collection Point
Haberl, Neuhold, etal. mobil.TUM 2017, to be published
![Page 27: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/27.jpg)
27
27
Results of Typical Motorway Segments – 2+1 Lanes: SAE0
5% SV, v80
5% SV, v100
5% SV, v130
15% SV, v80
15% SV, v100
15% SV, v130
10% SV, v80
10% SV, v100
10% SV, v130
No Results No Results No Results
No Results
Haberl, Neuhold, etal. mobil.TUM 2017, to be published
![Page 28: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/28.jpg)
28
28
Results of Typical Motorway Segments – 2+1 Lanes: SAE4 - 0.9seconds
5000m 300m 1200m
DCP 1 DCP 2 DCP 3 DCP 4
No Result!
DCP 1 DCP 2 DCP 3 DCP 4
+7.3% capacityof SAE0
+7.8% capacityof SAE0
+16.1% capacityof SAE0
DCP = Data Collection Point
Haberl, Neuhold, etal. mobil.TUM 2017, to be published
![Page 29: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/29.jpg)
29
29
Results of Typical Motorway Segments – 2+1 Lanes: SAE4 - 1.8seconds
5000m 300m 1200m
DCP 1 DCP 2 DCP 3 DCP 4
No Result!
DCP 1 DCP 2 DCP 3 DCP 4
+5.5% capacity of SAE0
+5.7% capacity of SAE0
+11.7% capacity of SAE0
DCP = Data Collection Point
Haberl, Neuhold, etal. mobil.TUM 2017, to be published
![Page 30: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/30.jpg)
30
30
Capacity Assessment via Speed-Flow-Diagrams without Breakdowns
1. Values considered: no increase in traffic volume over several time intervals
2. Determination of the 90th percentile of the values
3. Find the value corresponding to or closest to the 90th percentile
Haberl, Neuhold, etal. mobil.TUM 2017, to be published
![Page 31: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/31.jpg)
31
31
Results of Typical Motorway Segments – 2+1 Lanes
SAE LevelCapacity [veh/h]
DCP 1Capacity [veh/h]
DCP 2Capacity [veh/h]
DCP 3Capacity [veh/h]
DCP 4
SAE 0 3 015 2 790 3 270 3 063
SAE 4 - 0.9 sec NaN 2 990 (+7.2%) 3 526 (+7.8%) 3 556 (+16.1%)
SAE 4 – 1.8 sec NaN 2 953 (+5.5%) 3 472 (+5.7%) 3 478 (+11.7%)
5000m 300m 1200m
DCP 1 DCP 2 DCP 3 DCP 4
Mean speed [km/h], motorway segment - 2+1 lanes
DCP = Data Collection Point
+23% mean speedHaberl, Neuhold, etal. mobil.TUM 2017, to be published
![Page 32: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/32.jpg)
32
32
Conclusions of VEGAS
Summary
� Similar acceptance of automated vehicles with headway of 0.9 or 1.8 sec
� Capacity increase depends on position of data collection point
� Performance analysis on entire
motorway network (upcoming)
Methodological shortcomings
� Modeling of AD vehicles by driver.DLL to be improved
� unnecessary lane changes
� difficulties in lane keeping
� platooning to be developed
LOS A LOS B LOS C LOS D LOS E LOS F
Volume/Capacity ≤ 0,30 ≤ 0,55 ≤ 0,75 ≤ 0,90 ≤ 1,00 ≥1,00
Street segments 4 147 5 749 3 066 1 105 541 1 338
Ratio of LOS 26.0% 36.1% 19.2% 6.9% 3.4% 8.4%
![Page 33: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/33.jpg)
33
33
Lateral movement
Vissim contains years of experience in modeling lateral movement
1. DVU´s (driver-vehicle-units) push harder to neighboring lane as lane drop appears
2. Rear vehicles may open up gap
3. DVU´s may accelerate above desired speed
4. DVU´s do path planning by reverse routing
tree
![Page 34: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/34.jpg)
34
34
Example path planning in microsimulation
� Vissim network topology supports variety of infrastructure setups
• parallel links no lane change
• single multi-lane links lane change
![Page 35: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/35.jpg)
35
35
Path planning, urban arterial
� quick merge at both main lanes without usage of left turn bay
� similar to mainline direction (moderate acceleration)
� similar headway to lead and lag vehicle
![Page 36: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/36.jpg)
36
36
Path planning, urban arterial
� midpoint lane change crossing both lanes
a) accelerate quickly to cut rear vehicle
b) accelerate moderately to change behind rear veh.
a
![Page 37: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/37.jpg)
37
37
Path planning, urban arterial
� midpoint lane change crossing both lanes
a) accelerate quickly to cut rear vehicle
b) accelerate moderately to change behind rear veh.
b
![Page 38: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/38.jpg)
38
38
Path planning, urban arterial
� lateral movement lane-by-lane
a) moderate acceleration to change prior to lag vehicle
b) moderate deceleration to change behind lag vehicle
b
![Page 39: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/39.jpg)
39
39
TU Delft (van Lint, Hoogendorn)
� Delft integrated Traffic & Travel Laboratory comprises several research projects
(model applications and model development)
� Urban Mobility Lab with multimodal real-
time modeling of Amsterdam region
� Open Traffic Simulation (OTS)Many of the commercial software packages available have
their origins in the academic world, but are now developed and distributed by specialized software companies. In an
ideal world academia and practice collaborate continuously
along the lines sketched in Figure 1. In the real world, however, there is much fewer interaction between academia
and practice once simulation software is commercialized.
Particularly microscopic traffic simulation packages are black
boxes, in which it is impossible to unravel which underlying mathematical and numerical choices have been made, and
what the consequences of these are in terms of model validity.
This black-box problem essentially forces most academic
researchers to reinvent the wheel and to code an entire simulation model from scratch, even in case the research
objective is to develop a very specific innovation (e.g. a new
car following model).
![Page 40: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/40.jpg)
40
40
human factors (HF) in traffic flow simulation
� Car-following model: Human Driver Model (HDM) based on the former Intelligent Driver Model (IDM) by Treiber etal. and implemented in OpenTrafficSim (OTS)
� idea HDM: drivers have slow reaction time but anticipate future traffic states
Treiber, M., A. Kesting, and D. Helbing, Delays, inaccuracies and anticipation in microscopic traffic models. Physica A: Statistical Mechanics and its Applications, 2006. 360(1): p. 71-88
Van Lint, J. W. C., W. Schakel, G. Tamminga, P. Knoppers and A. Verbraeck (2016). Getting the human factor into traffic flow models – a new open-source design to simulate next-generation traffic operations. TRB (16-2917)
comfortable deceleration
reaction time, driver anticipates const accel/decel
stopping distance
desired, speed depend distance
� idea OTS: if drivers get distracted, desired speed drops and reaction time increases
![Page 41: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/41.jpg)
41
41
Example Human Factors
� at large reaction times deteriorates safety (collisions and time-to-collision (TTC))
Van Lint, etal., TRB (16-2917)
incident on opposite link leads to
max distraction at x2 keeping until x3
with gradual increase from x1
� predicting effects on efficiency and safety of vehicle automation
![Page 42: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/42.jpg)
42
42
MAVEN - Managing automated vehicle at signalized intersections
� Road infrastructure will monitor and support vehicles movements at signalized intersections
� Beyond ADAS and C-ITS with signal plan negotiation for adaptive traffic signal control
� develop suitable communication protocols, test via simulation and a real-world prototype (main contributions by MAP (NL), DLR, Hyundai, TomTom and city councils)
� particular infra-assisted signal control measures:
• phase negotiation (signal timing vs. arrival pattern), speed advisory, lane advisory
• Platoon management: forming, joining, progression, leaving platoons
• Conventional traffic and vulnerable road users
MAVEN is funded by the EC Horizon 2020 Programme under Grant No 690727, 09/2016-08/2019
![Page 43: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/43.jpg)
43
43
MAVEN – system concept
![Page 44: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/44.jpg)
44
44
concluding remarks
1. (Mobility-) times are changing rapidly; Connected & Automated Vehicles require numerical modeling
2. Microscopic traffic flow simulators within co-simulation environments
3. Software tools without programming interfaces not being applied in R&D
4. VISSIM (and others) have hidden mathematical models which could be useful for CAV-development
![Page 45: How traffic flow simulation can teach automated · PDF fileHow traffic flow simulation can teach automated driving ... simple calculation Vehicles need space: ℓ velocity, headway,](https://reader035.vdocuments.us/reader035/viewer/2022062600/5aa8bbaf7f8b9a81188be589/html5/thumbnails/45.jpg)
45
45
concluding remarks
1. (Mobility-) times are changing rapidly; Connected & Automated Vehicles require numerical modeling
2. Microscopic traffic flow simulators within co-simulation environments
3. Software tools without programming interfaces not being applied in R&D
4. VISSIM (and others) have hidden mathematical models which could be useful for CAV-development
Thank for your [email protected]