CADRE • NAV08 Presentation November 2008 Tim Thornton Smartcom Software tt@smartcomsoftware.com CADRE NAV08 Slide 1
Congestion CADRE Avoidance Dynamic Routing Engine Vehicle routing and traffic speed models for the real world Slide 2 NAV08
CADRE System Overview Slide 3 NAV08
CADRE The Congestion Problem Slide 4 NAV08
CADRE How To Tackle Congestion Build more roads Reduce use of vehicles on the roads More use of roads at low congestion times More use of uncongested roads CADRE Travel time predictions at different times Better route selection Dynamic vehicle routing Slide 5 NAV08
CADRE Overview Roadside GPS sensors vehicle data Mobile phone data Speed reports Road Speed network limits Speed model Route requests Speed Routing Speed engine engine data Routes Road network Slide 6 NAV08
CADRE Architecture Routing Data Client Logger Routing Journey Data Information Snapping Map Routing Engine Server Engine Speed Engine Short Term Model Long Term Model Road Journey Network Data Specific Road Generic Road Data Model Data Model Data Slide 7 NAV08
CADRE Journey Data Mobile Phone Reject Bad Data Ingest or Logger Data Reject Add to Data Unsnappable Snap to Roads Store Data Slide 8 NAV08
CADRE Snapped Data Snapping Development (since improved – about 90% pass rate) Green OK, Blue failed, since improved, Gold other journeys. Slide 9 NAV08
CADRE Speed Engine Scales Slide 10 NAV08
CADRE Long Term Speed Engine Core long term model in place Provides speeds for each road segment varying by time of day and day of week Long term model in process of improvement Short term model being developed in Matlab Slide 11 NAV08
CADRE Generic Speed Model For roads where there is no specific model Spread of data currently being reduced by better modelling Model training also being enhanced and automated A road dual carriageway, 70mph speed limit, Monday Slide 12 NAV08
CADRE Arc Specific Model A model is created for a specific arc once there is enough data of good enough quality The model is based on the generic one for that arc and then recalculated using the data for that arc Slide 13 NAV08
CADRE Short Term Model Aim is to detect congestion from live traffic data, and then create a short term predictive model based on that This part of the speed model is still in the initial modelling phase Speed LTM STM Data Time of day Now Slide 14 NAV08
CADRE Routing Engine Minimum time or minimum distance route Uses speed limit or CADRE speed model or speed model capped by speed limit Optimised for time varying road speeds Heuristics, such as pruning, to aid performance Always finds the true optimal route Slide 15 NAV08
CADRE Routing Engine Slide 16 NAV08
CADRE Hindhead Traffic Lights Journey time using speed engine is more accurate than using speed limit data or via Michelin Routing shows that in this case it pays to stay on the congested main road for almost all of the time, rather than to divert down narrow country lanes Slide 17 NAV08
CADRE Routing Client Slide 18 NAV08
CADRE Sample Results ‐ Hindhead Minimum Time – Speed Limit Minimum Distance Slide 19 NAV08
CADRE Sample Results ‐ Hindhead Minimum Time – Minimum Time – Speed Speed Engine – 12.30 Engine – 08.00 Slide 20 NAV08
CADRE Results Comparison Hindhead 12.30 Red distance, Blue time Slide 21 NAV08
CADRE • NAV08 Presentation November 2008 CADRE NAV08 Slide 22
Recommend
More recommend