Use a computer model and growth projections to simulate future demands on the transit network. Use a multi-objective optimizer (e.g. NSGA-III) to find the best transit options where the optimization functions: (1) reduce the average commute time, (2) minimize the cost to implement, (3) minimize the time to implement, (4) maximizes safety of the system, (5) minimizes standard deviation in commute time due to unforeseeable accidents, maintenance, etc. The model would have to be stochastic in nature to account for these real world happenings. The different transit options available to the optimizer include metro, bus, dedicated lanes, smart traffic light systems, HOV networks, dynamic highways, time-sensitive lanes, and even promoting telework stations. Let the optimizer present us with good choices and sort through all these variables!

### Model out the tradeoffs and talk about them

Tags: optimization traffic control tradeoff

Who would have the expertise to construct such a model, and would it be feasible to construct in a short period of time?

In addition, where could we get sufficient (complete and accurate) ground truth data to tune/train the model of the world? Without verifying the model, the optimizer's results may be no better than any other wild speculation.

0Vote up0Vote downAny number of beltway bandit contractors could construct such a model within a short period of time given the right data. And the data has to come from the government or private organizations ($$). A properly crafted optimizer's results considers all variables, all options that are fed into it, in all combinations, and can account for future growth. No other solution is this comprehensive. You could literally crowd-source the options to feed into it. Someone posted extending an exit ramp off I-66 to Route 27 -- estimate the cost and safety issues and add that as an option to the model. All options are mathematically considered to reduce commute time.

0Vote up0Vote down