From: I have a bad feeling about this http://t.co/aA81XJAk — newsycombinator (@newsycombinator)

Obtaining the maximum vehicle flow for each intersection can then be done with the following SQL statement:

select int_no, date, max(ND+NTD+NG+ED+ETD+EG+SD+STD+SG+OD+OTD+OG) as maxVehicleCount
from comptages group by idDemande, date

All comments welcome.

Here is what happens instead: for a given project, a researcher will obtain a copy of a dataset from a public organization, say traffic volumes, import this data in one’s own system, organize it, do some analysis, and that’s it. There is probably some thought given to the data organization for the specific goal, and the dataset was probably expanded one way or another for the complete analysis. During this time, the public organization went its own way, collecting more data and putting it (or not) in its own system, maybe even doing modifications or treatments (cleanup) to the data previously shared. So now, you have two diverging datasets with added value, typically in different formats, that will never be reconciled. Next project, it will start all over again. And voilà! Your tax dollars at work…

Yes, everyone is busy, everyone has deadlines, but what kills me is that when you hint at the problem and offer people to actually try to do something about it, think for a second and try to bring some consistency, your are quickly dismissed as an idealist. “That would be nice, but we have work to do”.But that, organizing datasets and agreeing on data exchange formats and protocols, is the work!

That is why I support open standards and data: choose an open format, clean the data minimally and make it public. Actually engage with the community, find use cases, and iterate to find an data organization/format that would meet most needs. Montréal has made some progress, but there is still a long way to go, and public practices, including academic practices, are unfortunately part of the problem.

“much of the existing ways in which knowledge is written down and maintained — from journals and peer review to libraries and copyright — is driven by the needs of a world based on paper”

• The New Einsteins Will Be Scientists Who Share

“This document is a testament to the contributions of AI and advanced computing techniques to transportation and there is every reason to think that this will continue. In particular, systems relying on these techniques should become more and more autonomous. As authors have noted repeatedly in this circular, a major disadvantage of current techniques is the selection of parameters and the lack of guidance, which leads to painstaking and time-consuming manual trial and error and sub-optimal operation. Truly autonomous systems will minimize user input and supervision by being able to adapt automatically to various environments and conditions.

A perhaps overlooked and significant force in this progress is the Internet and its enabling power: it enables in particular at a minimal cost the sharing of knowledge and collaboration in spatially distributed network of individuals. The use of AI and advanced computing techniques has become more widespread because many tools are freely available, many under open source licenses, because they can be re-used and modified which allows a better and faster bootstrapping of technology. Demonstrating the principles of swarm intelligence, re-use, sharing and collaboration allow human society to become more intelligent and better equipped to adapt to a changing environment and to cope with the complexity and challenges of ensuring safe and fast transportation for a growing population.”

Anyway, we just got our first paper accepted in Accident Analysis & Prevention, which was a long, long, time coming up. It is an analysis of the results of my PhD (yes, 2005…) that was not easy to put together as it addresses exposure to the risk of collision, but does not provide direct measure of safety, and is a long distance collaboration with my former supervisor Sophie Midenet at IFSTTAR (previously INRETS). I’ll let you with the abstract:

This paper proposes an original definition of the exposure to lateral collision in signalized intersections and discusses the results of a real world experiment. This exposure is defined as the duration of situations where the stream that is given the right-of-way goes through the conflict zone while road users are waiting in the cross-traffic approach. This measure, obtained from video sensors, makes it possible to compare different operating conditions such as different traffic signal strategies. The data from a real world experiment is used, where the adaptive real-time strategy CRONOS (ContRol Of Networks by Optimization of Switchovers) and a time-plan strategy with vehicle-actuated ranges alternately controlled an isolated intersection near Paris. Hourly samples with similar traffic volumes are compared and the exposure to lateral collision is different in various areas of the intersection and various traffic conditions for the two strategies. The total exposure under peak hour traffic conditions drops by roughly 5 min/h with the CRONOS strategy compared to the time-plan strategy, which occurs mostly on entry streams. The results are analyzed through the decomposition of cycles in phase sequences and recommendations are made for traffic control strategies.

Midenet, S.; Saunier, N. & Boillot, F. Exposure to Lateral Collision in Signalized Intersections with Protected Left Turn under Different Traffic Control Strategies Accident Analysis & Prevention, 2011 (preprint)

Via The Transportationist