© ComTec 2017 1

Contact

Prof. Dr.-Ing. Klaus David

Kassel UniversityFaculty of Electrical Engineering / Computer ScienceChair for Communication Technology (ComTec)Wilhelmshöher Allee 7334121 Kassel - Germany

Phone: +49 - 561 - 804 - 6314Phone2: +49 - 561 - 804 - 6446 (Secretary)Fax: +49 - 561 - 804 - 6360Mobile: +49 - 170 - 2901602Email: david@uni-kassel.de

WWW: http://www.comtec.eecs.uni-kassel.de/

Chair for Communication Technology

Prof. Dr.-Ing. Klaus David

© ComTec 2017 2

Pedestrian safety as a “killer”

Application for 5G:

The Wireless Seat Belt:

Prof. Dr. Klaus David

© ComTec 2017 3

Acknowledgement

• To the following Consultants/ Researchers/ PhD

students– Dr. Alexander Flach – now with OEM supplier

– Sebastian Engel – now with OEM

– Michel Morold

– Marek Bachmann

– Christoph Anderson

– Dr. Isabel Hübener

– Dr. Hendrik Berndt

– Prof. Dr. Lau, Sunvay University, Malaysia

– Prof. Dr. Stephan Sigg, Aalto University, Helsinki, Finland

– And more from the ComTec Team

© ComTec 2017 4

Overview

• Introduction

• Current Solutions and accident scenarios

• “Ideal” System?

• Various requirements

• Conclusion

© ComTec 2017 5

Overview

• Introduction

• Current Solutions and accident scenarios

• “Ideal” System?

• Various requirements

• Conclusion

© ComTec 2017 6

Introduction

• According to the latest report on road safety of

the Word Health Organization (WHO) [1],

pedestrians comprise 22% of all road traffic

deaths, approximately 275,000 worldwide

[1] Global status report on road safety 2015. Geneva, Switzerland: World Health Organization, 2015.

© ComTec 2017 7

Introduction

• For Germany 2015:– 3459 killed in traffic accidents

– 15,5 % (537) pedestrians

– while 11,1% (383) involved bicyclistsSource:https://www.destatis.de/DE/Publikationen/Thematisch/TransportVerkehr/Verkehrsunfaelle/VerkehrsunfaelleJ2080700167004.pdf?__blob=publicationFile

© ComTec 2017 8

Overview

• Introduction

• Current Solutions and accident scenarios

• “Ideal” System?

• Various requirements

• Conclusion

© ComTec 2017 9

Passive Approaches

• Optimized design of the car, so that collisions

harm pedestrians less.

© ComTec 2017 10

Passive Approaches

• Automatic opening and lifting of the bonnet (hood)

• Suspensions of the windscreen wiper hidden

under the front part

• Concepts of outside the car “air-bags”

Source: www.volvo.com

© ComTec 2017 11

Active Approaches

• Radar

• Cameras

• Infrared

• LIDAR (Light Detection and Ranging)

• Already available in products

• Some disadvantages:

– Not 100% reliable (snow, darkness, rain, obfuscations)

– Typically work “only” once the pedestrian is already on

the road and only in situations without obfuscations

Source: www.volvo.com

© ComTec 2017 12

Accident Scenarios

Statistics from the GIDAS (German In Depth Accident Study) database

© ComTec 2017 13

Accident Scenarios

⇒ About 80% rectangular crossing of street

(30% with obfuscation)

Published in: [2] S. Engel, C. Kratzsch, and K. David, “Car2Pedestrian-

Communication: Protection of Vulnerable Road Users Using Smartphones,” in 17th

Int. Forum Advanced Microsystems for Automotive Applications (AMAA 2013) Berlin,

Germany, 2013.

© ComTec 2017 14

Overview

• Introduction

• Current Solutions and accident scenarios

• “Ideal” System?

• Various requirements

• Conclusion

© ComTec 2017 15

© ComTec 2017 16

Is there a more ideal solution?

© ComTec 2017 17

© ComTec 2017 18

Characteristics of an “Ideal” System

• Close to optimum possible solution!

• What is needed?:

– Complete overview of the scenario (physics of the car

and of the pedestrian)

– Movement history of the pedestrian

– Personalized information about the pedestrian

(weight, child/ elderly person, and many more)

– Filtering (Context Filter!) of the pedestrian in danger

– “Fast” Exchange of few data between Car and P

© ComTec 2017 19

© ComTec 2017 20

ATTENTION PEDESTRIAN!

1

2

2

3

3

Direct Communication

Cellular Communication

Central Server

1

2

ATTENTIONCAR!

© ComTec 2017 21

Short Literature Overview

© ComTec 2017 22

Source

https://eandt.theiet.org/content/articles/20

17/10/smart-road-crossing-warns-drivers-

about-phone-fixated-pedestrians/

Last checked: 9.10.17

Alternatives

© ComTec 2017 23

Literature

• C. Sugimoto, Y. Nakamura, and T. Hashimoto,

“Development of Pedestrian-to-Vehicle

Communication System Prototype for Pedestrian

Safety Using both Wide-Area and Direct

Communication,” in IEEE 22nd International

Conference on Advanced Information

Networking and Applications (AINA) Ginowan,

Japan, Mar. 2008, pp. 64–69.

• NTT Docomo, the University of Tokio

© ComTec 2017 24

Publications:

• A. Flach and K. David, ”A Physical Analysis of an Accident Scenario between Cars and Pedestrians “, IEEE VTC, Anchorage, Alaska, USA, 20 – 23 September, 2009

• K. David, A. Flach: “CAR-2-X and Pedestrian Safety”, in IEEE Vehicular Technology Magazine, IEEE, Volume: 5 Issue:1, 2010, pp70-76, March 2010

• A. Flach, K. David, “Combining Radio Transmission with Filters for Pedestrian Safety: Experiments and Simulations”, 6–9 September 2010, IEEE VTC Fall 2010, Ottawa, Canada

• A. Flach, A.Q. Memon, “Pedestrian Movement Recognition for Radio Based Collision Avoidance: A Performance Analysis”, accepted for presentation on IEEE VTC Spring 2011 Budapest

• C. Voigtmann, S.L. Lau, and K. David, "Evaluation of a collaborative-based filter technique to proactively detect pedestrians at risk", VTC2012-Fall, Quebec City, Canada, 3 - 6 Sept., 2012

• Schulz, A. Roßnagel, K. David, "Datenschutz bei kommunizierenden Assistenzsystemen im Straßenverkehr", ZD (Zeitschrift für Datenschutz), November 2012, pp. 510 - 515

• I. König, A. Q. Memon, and K. David. "Energy consumption of the sensors of Smartphones", IEEE ISWCS 2013, Ilmenau, Germany, 27 – 30 August, 2013

• A. Flach, and K. David, “Car and Pedestrian Collisions: Causes and Avoidence Techniques", in: Rola Naja (Editor), Wireless Vehicular Networks for Car Collisioin Avoidance”, Springer, 2013

• A. Q. Memon, S.L. Lau, and K. David. "Investigation and Compensation of the Magnetic Deviation on a Magnetometer of a Smartphone caused by a Vehicle", IEEE VTC 2013, Las Vegas, USA, 2 -5 Sept. 2013

• S. Engel (Audi), C. Kratzsch (Audi), and K. David, "Car2Pedestrian-Communication: Protection of vulnerable road users using smartphones", to be presented at 17th International Forum on Advanced Microsystems for Automotive Applications (AMAA 2013) Smart Systems for Safe and Green Vehicles, Berlin , Germany, 17-18 June 2013

© ComTec 2017 25

Literature

• X. Wu et al, “Cars Talk to Phones: A DSRC

Based Vehicle-Pedestrian Safety System,” in

IEEE 80th Vehicular Technology Conference

(VTC2014-Fall) Vancouver, Canada, 2014, pp.

1–7.

• Honda, Qualcomm

• BMW, Bosch, with German Universities

• Audi, University of Kassel

© ComTec 2017 26

Overview

• Introduction

• Current Solutions and accident scenarios

• “Ideal” System?

• Various requirements

• Conclusion

© ComTec 2017 27

What about timing requirements?

• A. Flach and K. David, ”A Physical Analysis of an

Accident Scenario between Cars and

Pedestrians “, IEEE VTC, Anchorage, Alaska,

USA, 20 – 23 September, 2009

© ComTec 2017 28

Physical analysis of an accident scenario

• time available tsta for a collision avoidance system

– Influenced by

• trea (fixed) = Reaction + break response time = 0.83s

• tbra = Braking time (influenced by speed / deceleration)

• Communication radius

© ComTec 2017 29

0

1

2

3

4

5

6

7

8

9

10

1 2 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

• 1. Quadratically reduction tsta due to the increased

braking distance

• 2. Higher speed, less time needed to cover remaining

distance reducing tsta

Dependency between system time

available tsta and speed vcar

tsta [s]

vcar [km/h]

Scom = 50mScom = 50m

Decelaration: 10.4 m/s²

© ComTec 2017 30

Example for physical parameters

System time available for parameter values of scenario:

• vcar = 50km/h (about 31 miles/hour)

• scom= 50m

• trea = 0.83s

• Deceleration = 10.4m/s²

tsta= 2.1s

© ComTec 2017 31

What about position requirements?

© ComTec 2017 32

Source: https://spectrum.ieee.org/tech-talk/semiconductors/design/superaccurate-gps-chips-coming-to-smartphones-in-2018Last checked: 9.10.17

© ComTec 2017 33

Source: https://spectrum.ieee.org/tech-talk/semiconductors/design/superaccurate-gps-chips-coming-to-smartphones-in-2018Last checked: 9.10.17

Those days are nearly at an

end. At the ION GNSS+

conference in Portland, Ore.,

today Broadcom announced

that it is sampling the first

mass-market chip that can take

advantage of a new breed of

global navigation satellite

signals and will give the next

generation of smartphones 30-

centimeter accuracy instead of

today’s 5 meters. Even better,

the chip works in a city’s

concrete canyons, and it

consumes half the power of

today’s generation of chips.

The chip, the BCM47755, has

been included in the design of

some smartphones slated for

release in 2018, but Broadcom

would not reveal which.

© ComTec 2017 34

What about accuracy requirements?

• M. Bachmann, M. Morold, and K. David,

“Improving smartphone based collision

avoidance by using pedestrian context

information,” WIP Session IEEE PerCom, Kona,

HI, USA, 2017, pp. 1–4.

© ComTec 2017 35

Methodology

• Evaluate the probability of a Missed Alarm (𝑃𝑀𝐴)

– Which is the probability that a impending collision is

not detected, depending on the accuracy of the

smartphone‘s sensor data

© ComTec 2017 36

Methodology: Evaluation Scenario

w2=0.2 m

w4=0.4 m

50 m

curb

w3=2.1 m

w1=1.3 m

w5=1.0 m

parked car

driving car

! "

±! $

±! %±! %

−! '

! '

−! "x [m]

y[m]

© ComTec 2017 37

Influence of Sensor Accuracy on 𝑷𝑴𝑨

10 20

30 40

50 60

0 0.2

0.4 0.6

0.8 1

0

0.2

0.4

0.6

0.8

1

PM

A

sFsV

PM

A

0

0.2

0.4

0.6

0.8

1

𝜎𝑋 = 𝜎𝑌 = 0.5𝑚

10 20

30 40

50 60

0 0.2

0.4 0.6

0.8 1

0

0.2

0.4

0.6

0.8

1

PM

A

sFsV

PM

A

0

0.2

0.4

0.6

0.8

1

𝜎𝑋 = 𝜎𝑌 = 1𝑚

10 20

30 40

50 60

0 0.2

0.4 0.6

0.8 1

0

0.2

0.4

0.6

0.8

1

PM

A

sFsV

PM

A

0

0.2

0.4

0.6

0.8

1

𝜎𝑋 = 𝜎𝑌 = 2𝑚

10 20

30 40

50 60

0 0.2

0.4 0.6

0.8 1

0

0.2

0.4

0.6

0.8

1

PM

A

sFsV

PM

A

0

0.2

0.4

0.6

0.8

1

𝜎𝑋 = 𝜎𝑌 = 4𝑚

© ComTec 2017 38

Concluding requirements

• For 𝑃𝑀𝐴 < 0.2 we need:

• 𝜎𝑋 = 𝜎𝑌 < 0.5𝑚

• V < 0.2 m/s

• and the inaccuracy of the direction < 15 °

© ComTec 2017 39

How can Machine Learning help?

• A. Jahn, K. David, and S. Engel, “ 5G / LTE

based protection of vulnerable road users:

Detection of crossing a curb", IEEE VTC,

Boston, USA, 7 - 9 Sept. 2015, pp. 1-5.

• M. Bachmann, M. Morold, and K. David,

“Improving smartphone based collision

avoidance by using pedestrian context

information,” WIP Session IEEE PerCom, Kona,

HI, USA, 2017, pp. 1–4.

© ComTec 2017 40

Detecting of crossing a curb

curb height

walking speed

• Walking speeds: – 0 km/h, 3 km/h, 6 km/h, 10 km/h

– 0┐0, 0┐3, 3┐3, 6┐6, 10┐10

• Curb heights:– 3 cm, 10 cm, 16 cm

© ComTec 2017 41

Methodology

• Sensor data recording

– Smartphone right pant pocket

– Accelerometer, Gyroscope

• Pre-Processing: feature extraction

• Activity model generation: KNN

• Evaluation: 10-fold cross-validation

Sensor data recording

Data pre-processing

Activity model

generation

model evaluation

© ComTec 2017 42

• Up to 80% recognition rate

• (3 km/h / 3 km/h, KNN and 10 cm)

© ComTec 2017 43

Curb Detection Scenario

Approaching Car

curb

© ComTec 2017 44

Curb Detection Scenario

Approaching Car

curb

© ComTec 2017 45

Curb Detection Scenario

Approaching Car

curb

© ComTec 2017 46

Curb Detection Scenario

Approaching Car

curb

© ComTec 2017 47

Curb Detection Scenario

Approaching Car

curb

© ComTec 2017 48

Curb Detection Scenario

Approaching Car

curb

© ComTec 2017 49

Overview

• Introduction

• Current Solutions and accident scenarios

• “Ideal” System?

• Various requirements

• Conclusion

© ComTec 2017 50

Open Research Questions

• Optimum Protocols and Architecture for D2D

and Cellular Communications

• How to “guarantee” privacy?

• Optimum Architecture for very low latencies

(Tactile Internet)

• Improvements of GPS (GNSS), or may be

localization within 5G

• ….

© ComTec 2017 51

Conclusion

• Pedestrian Safety is an important Challenge!

• Various passive and active approaches

• An “Ideal Solution” is possible and has been

presented here: The “Wireless Seat Belt”

• 5G with low latency and direct communication

would be an ideal network to realize this