CYCLIST BEHAVIOR AT DISCONTINUITIES IN THE CYCLING NETWORK Matin S. Nabavi Niaki, École Polytechnique de Montréal Nicolas Saunier, École Polytechnique de Montréal Luis Miranda Moreno, McGill University
Introduction Transportation engineers and planners are focusing on improving: cycling mode share cyclist safety The World Health Organization reported that about 1.25 million people die each year as a result of road traffic crashes (1) half of which are vulnerable road users In Canada, 3.2 % of all traffic fatalities involved cyclists (2) 1.3 % of Canadian commuters are cycling to work (2) 2
Introduction To improve safety and cyclist mode share, cyclist behavior must be studied throughout the cycling network Many studies have identified some of the reasons for low cycling mode share to be the low objective and perceived safety of cyclists which could be due to the discontinuities in the cycling network (6, 7, 8, 9, 10) Therefore the road network must be designed so it can accommodate different road users safely in different conditions and situations by better understanding the risks and factors related to discontinuities 3
Introduction: Discontinuity in Network Cycling network provides connectivity between origin and destinations by means of cycling facilities: separate cycling facility bike lane shared/designated roadway off road etc. 4
Introduction: Discontinuity in Network Cyclists face interruptions in the cycling network: e.g. discontinuities Discontinuities include (10): end of a cycling facility change of side of a cycling facility change in cycling facility type intersections re routing due to construction change in pavement quality variation in motorized traffic volume on roads along bike facilities bus stops that cut off cyclists on the cycling facility parking spaces where vehicles cut off cyclists on the cycling facility Discontinuities have only recently been introduced as a measure of cycling network performance (10) 5
Background Discontinuity: end of cycling facility End of cycling facility at Chemin de la Côte Sainte Catherine and Avenue Villeneuve, Montreal, QC (Google street view) 6
Background Discontinuity: change in cycling facility side End of cycling facility at Saint Catherine Street West and Boulevard de Maisonneuve Ouest, Montreal, QC (Google street view) 7
Background Discontinuity: change in cycling facility type End of cycling facility at Boulevard Pierre Bernard and Rousseau Street, Montreal, QC (Google street view) 8
Objectives of Current Study Cyclist behavior and safety at discontinuities has not been studied in the past Understand the microscopic impacts of discontinuities in the cycling network in terms of behaviour and safety The methodology is based on video analysis comparing cyclist behaviour at sites with and without discontinuities (control) motion pattern learning surrogate measures of safety 9
Methodology 1. Identify points of discontinuity 2. Collect video data at discontinuity and control sites a) change in cycling facility side b) change in type of cycling facility 3. Extract and classify road user trajectories from video data 4. Analyse cyclist behaviour using motion pattern learning 5. Study various strategies adopted by cyclists when faced with discontinuities 6. Assess cyclist safety through speed and other surrogate measures of safety 10
Methodology Identify points of discontinuity in Montreal using methodology proposed by (10) End of cycling facilities and changes in cycling facility type in Montreal, QC 11
Methodology Collect video data at eight locations with four discontinuity and four control sites: four of the locations are presented here Location of two discontinuity and control sites selected for video data collection in Montreal, QC 12
Methodology Locations for video data collection: Change in cycling facility side 13
Methodology Locations for video data collection: Change in cycling facility type and End of cycling facility 14
Methodology Video analysis Feature tracking Feature grouping 454 B Road user classification Classified trajectories Cyclist behaviour analysis Trajectory clustering Motion patterns with high cyclist proportions Video analysis steps for obtaining road user type and trajectory 15
Methodology Motion pattern learning: clustering of the dataset into more homogeneous subsets using the longest common subsequence (LCSS) similarity measure Motion pattern learning results showing all road user maneuvers 16
Discussion of Results: Cyclist Behavior Cyclist maneuvers at site with discontinuity (left) and control site (right) 17
Cyclist Behavior at Discontinuity Cyclist maneuvers at site with discontinuity: Movements from Southwest to Northeast 18
Cyclist Behavior at Control Site Cyclist maneuvers at control site: Movements from Northeast to Southwest 19
Cyclist Behavior at Discontinuity Cyclist maneuvers at control site: Movements from Northeast to Southwest 20
Discussion of Results: Cyclist Behavior Turning left from separate cycling facility into designated roadway left: two way road right: one way road Cyclist maneuvers at side with discontinuity (left) and control site (right) 21
Discussion of Results: Cyclist Behavior Left: turning left from road into separate cycling facility Right: turning left from separate cycling facility to separate cycling facility Cyclist maneuvers at side with discontinuity (left) and control site (right) 22
Next Step Assess cyclist safety by: obtaining speed information of all road users from the collected video data for safety analysis obtain conflict measures between cyclist and other road users 23
Conclusion Although many studies have looked into cyclist behaviour in different situations and conditions, discontinuities have been overlooked The use of cyclist trajectory clustering provided valuable information on the microscopic maneuvers of cyclists More maneuvers were observed by cyclists at discontinuities Given the variability in cyclist maneuvers, vehicles and pedestrian face unexpected movements from cyclists Future work: analyzing more sites to draw stronger conclusions 24
References 1. World Health Organization. 2016. Road Traffic Injuries. Retrieved September 15, 2016 (http://www.who.int/mediacentre/factsheets/fs358/en/). 2. Transport Canada. 2013. Canadian Motor Vehicle Traffic Collision Statistics: 2013. Retrieved July 2, 2015 (http://www.tc.gc.ca/media/documents/roadsafety/cmvtcs2013_eng.pdf). 3. Nosal, T., and L. F. Miranda Moreno. 2014. The Effect of Weather on the Use of North American Bicycle Facilities: A Multi City Analysis Using Automatic Counts. Transportation Research Part A: Policy and Practice 66:213 25. 4. Yan, X., M. Ma, H. Huang, M. Abdel Aty, and C. Wu. 2011. Motor Vehicle Bicycle Crashes in Beijing: Irregular Maneuvers, Crash Patterns, and Injury Severity. Accident Analysis & Prevention 43(5):1751 58. Retrieved July 24, 2014 (http://www.ncbi.nlm.nih.gov/pubmed/21658503). 5. Pucher, J. A. 2000. Making Walking and Cycling Safer: Lessons from Europe. Transportation Quarterly 54:25 50. 6. Nabavi Niaki, M. S., N. Saunier, and L. F. Miranda Moreno. 2017. Analysis of Cyclist Behaviour at Cycling Network Discontinuities Using Computer Vision. Accepted for the Transportation Research Board 96th Annual Meeting. 7. Dill, J., and J. Gliebe. 2008. Understanding and Measuring Bicycling Behavior: A Focus on Travel Time and Route Choice. Portland, Oregon. Retrieved June 18, 2015. (http://scholar.google.co.uk/citations?view_op=view_citation&hl=en&user=gnyezcmaaaaj&citation_for_view=gnyezcm AAAAJ:UeHWp8X0CEIC). 8. Ehrgott, M., J. Y. T. Wang, A. Raith, and C. van Houtte. 2012. A Bi Objective Cyclist Route Choice Model. Transportation Research Part A: Policy and Practice 46(4):652 63. Retrieved May 31, 2015 (http://linkinghub.elsevier.com/retrieve/pii/s0965856411001844). 9. Garrard, J., G. Rose, and S. Kai Lo. 2008. Promoting Transportation Cycling for Women: The Role of Bicycle Infrastructure. Preventive medicine 46(1):55 59. Retrieved March 1, 2015 (http://www.ncbi.nlm.nih.gov/pubmed/17698185). 10. Mekuria, M. C., P. G. Furth, and H. Nixon. 2012. Low Stress Bicycling and Network Connectivity. 11. Nabavi Niaki, M. S., N. Saunier, and L. F. Miranda Moreno. 2016. A Methodology to Quantify Discontinuities in a Cycling Network Case Study in Montreal Boroughs. Transportation Research Board 95th Annual Meeting.
Questions? This research project is funded by the Fonds de Recherche du Québec Nature et Technologies (FRQNT)