Download presentation
Presentation is loading. Please wait.
Published byCharles Heath Modified over 8 years ago
1
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 1 Functionality analysis of Isokylä highway tunnel automatic incident detection system (AID) Sauli Pahlman & Juuso Kummala Finnish Road Administration {sauli.pahlman, juuso.kummala}@tiehallinto.fi
2
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 2 Background to the topic of the presentation - Traffic management on E18 road
3
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 3 E18 - Background A central element of the Nordic Triangle, which links the Nordic countries and their capitals to each other, Russia and central Europe Road connection from the ports of Turku and Naantali through the Helsinki Metropolitan area to the Vaalimaa border crossing and Russia
4
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 4 Development of E18 Target is a high-quality transport system with terminals, border crossings, services and environment, serving the whole society The goal is to construct the motorway connection Turku - Helsinki - Hamina by 2015 so that the sections Turku - Helsinki and that of Hamina are completed by 2008 and the section Helsinki - Hamina in 2015.
5
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 5 The functionality analysis
6
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 6 Technical overview About technical implementation 16 incident detection cameras (of which 5 cameras outside the tunnel) and 435 m of tunnel 5 different detection classes (opposite direction, stopped vehicle, smoke etc.) The Isokylä tunnel acts as a testbed for later (and longer) tunnel implementations like the ones on E18 (2.2 km, ~70 cameras) and Vuosaari port (1.6 km, ~55 cameras) The problem is that even the 16 cameras of Isokylä seem to generate a lot of false alarms -> too much for reliable use in TMC!
7
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 7 Description of the study 1300 video clips (30 s before and after the ”alarm”; gathered during winter and spring 2004/2005) were examined Verification of the alarms were done based on the videos It took 2 weeks to watch and verify the alarm videos Example videoclips shown later The motivation for the study was the large amount of false alarms reported by the FinnRA TMC operators Based on test drives it seems reasonable to assume that the number of undetected incidents is rather small (except in case of vehicle with alarm light; maintenance vehicles, police, rescue services etc.) The study is based on data collected before and after a firmware update (6.7.2005) which was supposed to improve system's performance
8
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 8 Overall quality of operation before 6.7.2005 Monthly distribution of alarms 12/2004 - 6/2005 The false alarms do not distribute equally, but they are clustered on certain days. These days tend to be very busy for TMC operators and risk of an indident is already higher than usually. For example during the winter storm at 22th december 2004 there were 70 alarms produced in 4 hours - all false.
9
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 9 Overall quality of operation after the update at 6.7.2005 Monthly distribution of alarms 8.7.2005 - 28.7.2005 An average of 3.5 false alarms per day.
10
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 10 Problems The system is not a real pattern recognition system but it is based on changes in areas (clusters of pixels in digitized video) of picture, thus the system can not distinquish between a shadow or reflection and a vehicle. Additionally, each recognition area is taken care by just one camera which is located above the traffic flow. Thus, the system gets no information about the dimensions of an object. Preceeding is the reason for example shadows, reflections and tyre marks to cause false alarms. The problems are clearly related to weather. On a rainy weather the road surface reflects light, during snowfall there is snow dropping from vehicles' roofs and during shiny weather the problem is shadows. False alarms are thus concentrated to points of time when conditions are already difficult (sunshine dazzles, rain degrades visibity).
11
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 11 Camera-specific functionality before the update The problems are centered around outside cameras and the tunnel cameras near to tunnel entrances. Outside cameras
12
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 12 Factors causing false alarms Most of the false alarms are caused by difficult weather conditions, namely rain and snowfall. Vehicles' headlights produce false alarms only in the cameras that scan the traffic flow from the front. Example clips 1.TruckTruck 2.SnowfallSnowfall 3.Wet roadWet road 4.Tyre marksTyre marks 5.Snow in the tunnelSnow in the tunnel 6.Head- lightsHead- lights 7.ShadowShadow 8.LightingLighting 9.CorrectCorrect
13
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 13 Measures made to improve AID's performance Setting up communication between AID and nearby weather stations so that outside cameras will not generate alarms while heavy snowfall will hopefully reduce the number of false alarms. Totally removing the outside cameras would significantly reduce the number of weather-related false alarms but the same time 1/3 of correct alarms. Alerting the TMC to "wrong direction" -alarm only if two consecutive cameras have generated alarms in short enough time scale. Blocking the "stopped vehicle" -alarm while the automatic lighting system is adjusting the its lighting level.
14
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 14 Discussion There are just a few competing systems available so there is no real competition. Video-based systems has an advantage because of easy and reliable verification However the market is growing and an increase in the number of actors in the market could improve the quality of AIDs' operation.
15
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 15 Epilogue E18 Muurla–Lohja motorway section When the new section between Lohja and Muurla has been completed, there will be a motorway from Turku all the way to Helsinki (and of course vice versa).
16
pp.kk.2003 / tekijän puumerkkiKalvosarjan nimi 16 The Lohja - Muurla motorway section in figures The project will be implemented using the life-cycle model: road construction and maintenance will be acquired with a single service agreement (25 years). Motorway: 51.3 km Ramps: 16.0 km Interchanges: 8 Bridge sites/bridges: 48/76 Double tunnels: 7/5.2 km (longest one 2.2 km) Groundwater protection: 4.2 km Construction period : 2005 - 2009 In service: : late 2008
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.