![Introduction Traffic Maps Prototypes Results Conclusions
• User Feedback
“I hated the [Google and equal-interval] maps
because you really had to think, and do math in
your head, which I did not enjoy.”
“The [fixed-minute] map was easiest...The other
maps were confusing.”
“I really hope they never start using the type of
maps that had 3 different colors for slow, fast, etc.
I‟d never get anywhere on time.”](https://image.slidesharecdn.com/stevensaag12-120227173408-phpapp02/85/Stevens-Goldsberry-Fixed-interval-Segmentation-for-Travel-time-Estimations-in-Traffic-Maps-34-320.jpg)
Stevens-Goldsberry | Fixed-interval Segmentation for Travel-time Estimations in Traffic Maps
- 1. Fixed-interval Segmentation for Travel-time Estimations in Traffic Maps JoshuaStevens KirkGoldsberry Feb. 25 | AAG 2012 NY Spatiotemporal Thinking, Computing, and Applications IV
- 2. Introduction Traffic Maps Prototypes Results Conclusions • Outline – Traffic, congestion, uncertainty – Maps as part of the solution – Trouble w/ existing designs – Prototypes: Considering segmentation – Empirical evaluation – Findings – Conclusions – Limitations
- 3. Introduction Traffic Maps Prototypes Results Conclusions • Congestion is a geographic hindrance, affecting millions of drivers every day
- 4. Introduction Traffic Maps Prototypes Results Conclusions • More than an inconvenience – Congestion = costs – 4.8 billion hours in delays – 1.9 billion gallons in wasted fuel – $101 billion in expenses paid by commuters 2011 Urban Mobility Report, Texas Transportation Institute
- 5. Introduction Traffic Maps Prototypes Results Conclusions • Congestion in other terms – $713 per urban commuter in 2010 – Experienced delay = 34 hours/yr • Or…4 vacation days
- 6. Introduction Traffic Maps Prototypes Results Conclusions • Systems exist to enhance efficiency – ITS: Intelligent Transportation Systems – ATIS: Advanced Traveler Information Systems • Usually broadcast by radio and in-vehicle units “ATIS-equipped drivers make better decisions and fewer late route diversions, which increases network efficiency.” - Al-Deek and Khattak (1998)
- 7. Introduction Traffic Maps Prototypes Results Conclusions • Decisions can be aided by travel-times Rand McNally Road Atlas
- 8. Introduction Traffic Maps Prototypes Results Conclusions
- 9. Introduction Traffic Maps Prototypes Results Conclusions
- 10. Introduction Traffic Maps Prototypes Results Conclusions
- 11. Introduction Traffic Maps Prototypes Results Conclusions • Variations of a single design approach
- 12. Introduction Traffic Maps Prototypes Results Conclusions • Typical traffic maps don‟t provide answers ….they pose questions • To determine travel-time: – Estimate distance(s) – Guess velocities – Perform mental arithmetic
- 13. Introduction Traffic Maps Prototypes Results Conclusions
- 14. Introduction Traffic Maps Prototypes Results Conclusions • Conventional traffic maps lack explicit segmentation – Multiple and varying segment lengths – Ambiguous velocities • We can address these limitations through map design
- 15. Introduction Traffic Maps Prototypes Results Conclusions • We designed two prototypes: – Improve distance estimation – Provide more detailed velocities – Or…represent travel-times directly
- 16. Introduction Traffic Maps Prototypes Results Conclusions • Prototype 1: Equal-interval method – Fixed segment lengths used elsewhere Source: USGS
- 17. Introduction Traffic Maps Prototypes Results Conclusions
- 18. Introduction Traffic Maps Prototypes Results Conclusions • Prototype 2: Fixed-minute method – Each segment has temporal length – Reminiscent of isochrone approach • But…not restricted to a single origin or destination
- 19. Introduction Traffic Maps Prototypes Results Conclusions
- 20. Introduction Traffic Maps Prototypes Results Conclusions • Can map-readers estimate travel-times? – Do our designs improve these estimations? • We designed an experiment – 60 questions over 3 categories • Arithmetic • Distance estimation (control vs segmented) • Travel-time estimation (conventional vs our proposed alternatives)
- 21. Introduction Traffic Maps Prototypes Results Conclusions • About our participants – n = 49 – Ages 18 – 33 (mode 20) – All had driving experience – All were MSU students, solicited across campus – Given $10 for participation
- 22. Introduction Traffic Maps Prototypes Results Conclusions • Arithmetic
- 23. Introduction Traffic Maps Prototypes Results Conclusions • Arithmetic – Considered „correct‟ if within 15% • 74.08 % correct, 87.14% confident 100 80 60 40 20 0 Correct Confident t (mean > n Min Max Median Mean % Error p threshold) 490 0.00 316.70 4.17 16.47 1.01 0.31
- 24. Introduction Traffic Maps Prototypes Results Conclusions • Distance Estimation
- 25. Introduction Traffic Maps Prototypes Results Conclusions • Distance Estimation
- 26. Introduction Traffic Maps Prototypes Results Conclusions • Distance Estimation
- 27. Introduction Traffic Maps Prototypes Results Conclusions • Distance Estimation
- 28. Introduction Traffic Maps Prototypes Results Conclusions • Travel-time Estimation
- 29. Introduction Traffic Maps Prototypes Results Conclusions • Travel-time Estimation
- 30. Introduction Traffic Maps Prototypes Results Conclusions • Travel-time Estimation
- 31. Introduction Traffic Maps Prototypes Results Conclusions • Travel-time Estimation % Correct Conventional 27.55% Equal-Interval 26.33% Fixed-minute 79.39% Mean Error ANOVA p-Matrix: Error Equal-interval Fixed-minute Conventional 19.77 minutes Fixed-minute 0.011 - Equal-Interval 13.57 minutes Fixed-minute 5.52 minutes Conventional 0.044 < .001
- 32. Introduction Traffic Maps Prototypes Results Conclusions • Travel-time Estimation Mean Response Time Conventional 26.03 s Equal-Interval 28.07 s Fixed-minute 18.64* s
- 33. Introduction Traffic Maps Prototypes Results Conclusions • User Preference
- 34. Introduction Traffic Maps Prototypes Results Conclusions • User Feedback “I hated the [Google and equal-interval] maps because you really had to think, and do math in your head, which I did not enjoy.” “The [fixed-minute] map was easiest...The other maps were confusing.” “I really hope they never start using the type of maps that had 3 different colors for slow, fast, etc. I‟d never get anywhere on time.”
- 35. Introduction Traffic Maps Prototypes Results Conclusions • Conclusions – Design matters: Segmentation can significantly influence estimation – Conventional designs are not sufficient • Most popular design makes conditions appear less severe than they are – These differences are likely to affect route decisions – Not limited to traffic maps • Great opportunities for rail/subway in particular
- 36. Introduction Traffic Maps Prototypes Results Conclusions • Limitations – Suitable for in-vehicle navigation units? • Oblique view not evaluated – Difficult symbology • Requires some manual adjustments • Presents challenge for real-time implementation – University students should be good w/ math • Same results with general sample?
- 37. Introduction Traffic Maps Prototypes Results Conclusions • Acknowledgements – My M.S. committee at MSU • Dr. Kirk Goldsberry, Dr. Judy Olson, Dr. Ashton Shortridge – California Dept. of Transportation – The MSU department of Geography – Many at Penn State for advice, critique, and new ideas
- 38. Introduction Traffic Maps Prototypes Results Conclusions • Questions & Contact Info Joshua Stevens | josh.stevens@psu.edu Kirk Goldsberry | kgoldsberry@fas.harvard.edu
Editor's Notes
- #4: Congestion introduces environment uncertainty…we know where we are going, but not how long it will take!
- #5: 4.8 billion hours >= 1,400 days of Americans playing Angry BirdsNew studies emerging linking traffic to heart disease and other illnesses (due to emissions, not rage!)Largest sectornot shipping and freight, but light-duty commuters
- #6: Recession = less money, less excess travel. Better ways to get people driving less -> make their drives more efficient
- #7: Signs along highway ‘tune into AM 410…”… but we can also map this info!
- #8: Based on static speed limit informationWe can improve this with real-time traffic conditions
- #14: Recall the conventional design. What does slow mean? Fast? Can you estimate route distance using linear scale bar?
- #17: Has precedent…sort of. Dashes used to establish hierarchy between road types.On 1:24k map, dash size is 240 feet, but this isn’t explicitly stated on map, nor designed for estimation.
- #24: Unreasonable to expect perfect accuracyArithmetic not a problem…poor travel-time estimations then relate to map interpretation
- #28: Clearly segmentation has a positive influence on distance estimation
- #32: Directional bias with conventional map (11.77 minutes underestimation) conditions look better than they areSlight over-estimation with fixed-minute map….91 seconds
- #33: Fixed-minute design the only significant difference p = .007).