Presentation is loading. Please wait.

Presentation is loading. Please wait.

IMPROVING INCIDENT DETECTION KPI ON SANRAL’S FREEWAYS IN GAUTENG

Similar presentations


Presentation on theme: "IMPROVING INCIDENT DETECTION KPI ON SANRAL’S FREEWAYS IN GAUTENG"— Presentation transcript:

1 IMPROVING INCIDENT DETECTION KPI ON SANRAL’S FREEWAYS IN GAUTENG
Claire Birungi & Avi Menon

2 Background An incident is any event or occurrence that can have an impact on normal traffic flow on the freeway. Incident Management Systems include Closed Circuit Television (CCTV) cameras, Variable Message Signs (VMS), Vehicle Detection Sensors (VDS) and incident response vehicles. The purpose of IMS is to: - Identify traffic incidents - Reduce incident occurrence in a proactive and reactive way - Provide emergency response to save lives and reduce delays

3 Incident Management Systems
CCTV Cameras Variable Message Signs On-Road Response Vehicles *Primary role is to identify incidents on the freeway *Proactive measure to mitigate incidents *Reactive measure to clear the scene and reduce delays

4 Detection of Incidents
Operators at the Traffic Management Centre (TMC) look for traffic incidents. Over 90% of the incidents are detected using CCTV. Detection is done by manually panning, tilting and zooming each camera. Incidents identified are recorded into the ATMS.

5 Occurrence Time & Detection Time KPI
Incident Occurrence Time is the actual time the incident occurred – obtained by rewinding video footages. Incident Detection Time is the time when the operator detects the incident using CCTV cameras and registers it on the ATMS. Incident Detection Time – Incident Occurrence Time < 3 minutes *Of the 3200 (on average per month) incidents detected in 2018, 70% do NOT have occurrence time *Limits ability to quickly respond to crashes *Decreases the efficiency of the whole network in terms of congestion and delays

6 Why Occurrence Time is Unkown
The operator rarely sees the incident occurring Camera facing away from incident occurred Camera zoomed in past the incident location Manually panning through the cameras Capturing the details of the previous incident in the ATMS.

7 Manual Surveillance Method
*Camera spacing varies from 400m to 1km No continuous sequential coverage *Continuously monitoring up to 32 cameras per shift The time of day, traffic volume on the section of road, known hotspot area and the number of operators available per shift.

8 Methodology Develop alternative methods of surveillance
- Less cumbersome - Improve coverage - Operator efficiency To increase number of incidents detected at the time of occurrence Test the method using a before-and-after study Before period: May 2017 & After period: May 2018 During the study, the TMOs, the TMC setup, Camera type and spacing remained as existing.

9 Static-Continuous View Surveillance
Position during the AM Position during the PM *Fixed position to view the entire freeway segment from end to end *Continuous and holistic view of the freeway segment.

10 Roaming View Surveillance
Works effectively where: - The distance between the cameras is short enough to provide an uninterrupted view of the freeway.

11 Automated Pre-set Surveillance
Freeway segments Interchange *Programming the camera to automatically zoom, tilt and pan to different ‘viewports’. *The cameras automatically cycle through these viewports allowing the operator to easily view the freeway segments

12 Comparison of the Surveillance Methods
Pros Cons Static- continuous view surveillance Continuous and holistic view of the freeway segment; Less labour intensive. Not feasible where distance is more than 400 meters Infill cameras Roaming view surveillance Continuous and quick view of the freeway; Holistic view of the roadway segment; Incidents on ramps may not be easily detected Automated pre- set surveillance View the segment or interchange within 3 minutes More than one pre-set cycle within 3 minutes. Continuous and quick view of the freeway Holistic view of the roadway segment Less labour intensive The setup of Pre-sets is time consuming

13 Automated Pre-set Surveillance
Implementation of pre-sets was conducted in March 2018 The camera is programmed to automatically view all sections of the freeway To determine viewports analysis of the following: - Traffic data analysis (volume, direction, peak period …) - Known accident hotspot areas - Camera spacing - Obstructions (overhead signboards, gantries …) - Geometry and length of the road Determine the time allocation for each viewport

14 Automated Pre-set Surveillance
CCTV Cameras Locations/Direction No of Monitored Freeway Segments (Viewports) Pre-set time for each freeway segment (seconds) Pre-set time to view all segments (seconds) No of Pre-set Tour Cycles in 3 minutes CAM 551 N17 Elands I/C EB AF N17 TO N17 Germiston I/C EB BF N17/Ramp to N3 1 3 12 CAM 501 15 N17 WB AF N17 Germiston I/C/Ramp from N3 TO N17 WB BF N17 Elands I/C 4 18 CAM 502 N17 EB AF N17 Germiston I/C/Ramp from N3 TO N17 Germiston I/C EB BF N17/Ramp from N3 5 2 10 9 CAM 327

15 Before & After Study May 2017 to May 2018 – incidents with occurrence increase by 500 incidents The increase was normalized with the total number of incidents for each month – resulting to and increase of 15%

16 Existing Shortcomings
System unavailability due to power outages Distance between cameras are higher than the ideal 400m Several blind spots on the network The setup of the operators – the same operator views camera footages on and also logs in data in the ATMS Bad weather conditions affect the visibility of the camera

17 Questions


Download ppt "IMPROVING INCIDENT DETECTION KPI ON SANRAL’S FREEWAYS IN GAUTENG"

Similar presentations


Ads by Google