Background: ICTCT Extra Workshop Vancouver 2018 Pedestrian workers - - PDF document

background
SMART_READER_LITE
LIVE PREVIEW

Background: ICTCT Extra Workshop Vancouver 2018 Pedestrian workers - - PDF document

Background: ICTCT Extra Workshop Vancouver 2018 Pedestrian workers have tasks: Which are carried out on foot on the road network (police officers, signalers, etc.) Who are on foot for short periods of time (facilitating traffic,


slide-1
SLIDE 1

1

Working in traffic as a pedestrian:

what are the risks for safety?

ICTCT Extra Workshop Vancouver 2018

Marie-Soleil Cloutier, INRS-UCS Nicolas Saunier, Polytechnique Montréal Sylvanie Godillon, INRS-UCS César Tartari, Polytechnique Montréal Daniel Lafond, Thales François Vachon, Université de Laval March 8-9 2018

2

Pedestrian workers have tasks:

  • Which are carried out on foot on the road network (police officers,

signalers, etc.)

  • Who are on foot for short periods of time (facilitating traffic,

making repairs, etc.)

  • Who frequently pass from their vehicle to the street (delivery

drivers). They are interacting with several road users: other workers (in vehicles), drivers, cyclists, pedestrians, residents, etc. They are exposed daily to high levels of stress and insecurity.

Background:

3

Road accident fatalities account for between 25% and 30% of all accidental work-related deaths (Pignatelli et al., 2013) In Quebec, Canada, between 2000 and 2008:

  • Pedestrian workers accounted for 6% of work accidents,
  • But for 23% of serious injuries and fatalities
  • And the highest median amount of compensation for road

accident victims at work (5 168 $ Vs. 1 704 $ for all) (Pignatelli et al., 2013)

Background:

4

Is there a link(s) between work environments and road accident risks for pedestrian workers?

Research question:

5

  • To study the working conditions of a specific group of

pedestrian workers: police officers doing traffic

  • To analyze the accident risk situations to which these

workers are exposed during their work with conflict methods

  • To explore their stress level (not presented here)

Objectives:

Police officer in Montreal and Quebec City

Methods

Recruitment of participants thanks to SPVM and SPVQ 19 participants:

  • Montreal: 8 police officers (3 men/5 women)
  • Québec: 11 police officers (10 men/1 woman)

6

slide-2
SLIDE 2

7

Telescopic pole

+

Digital video camera Solar mobile charger

+

During several hours A few days later… Tablet survey + Stress felt Interview Video recording Observation

Data collection (summer 2017)

Video analysis of the trajectories and interactions

Methods

Video analysis of the trajectories and interactions of road users is done automatically using a computer vision techniques. An interaction is formed by two users coexisting in the monitored area. Interactions between all users are analyzed according to various safety indicators, in particular:

  • Traffic conditions: speed and proximity
  • Time to collision (TTC): the time remaining before two users

collide if they continue on their current trajectories

  • use of motion prediction at constant velocity (preliminary)
  • post-encroachment time (PET): the time between the

departure of the first user and the arrival of the second user in the zone where their trajectories intersect

8

Video Analysis

  • Using the open source Traffic Intelligence

project

  • Data preparation

– metadata and camera parameters

9

correction for distortion

Video Analysis (cont.)

  • 1. Feature tracking
  • 2. Feature grouping
  • 3. Road user classification
  • 4. Interaction and indicator computation

10

1 2 grouping

11 11

frame 80

12 12

frame 100

slide-3
SLIDE 3

13

frame 120

14

frame 140

Sample Data: 4 sites, 2 workers

Workers\Sites Cluster 1 Cluster 2 P102 C110 28 min C109 30 min P108 C114 24 min C113 28 min

C110 C109 C114 C113

User Median Speed

16

Interaction Distance

17

TTCmin

18

Filter TTCs below 10 s

slide-4
SLIDE 4

PET

19

below 10 s

Conclusion

  • Work in progress
  • Large data collection effort requires a lot
  • f data organization and preparation

(ongoing)

  • Speeds are high and workers in the

middle of traffic have small TTC and PET with other road users

  • No link to site typology (for now)

20

Limitations

  • The worker is difficult to track

automatically

– static for long periods of time or off-duty on sidewalk

  • Unclear how accurately we can

automatically characterize the severity

  • f worker interactions in the middle of

traffic

21 22

Thanks for your listening ! And thanks to our interns who collected data :

  • L. Barrachina, F. Bataille, R. Jennings, Y. Ishimo, C.

Tartari

This presentation is based on

  • ngoing research funded by the

Robert Sauvé Research Institute on Occupational Health and Safety (IRSST).

TTC and PET

23

(Laureshyn 2010)