Open lmz opened 1 year ago
Comments from prior discussion:
SFCTA: I was super impressed by the work done for SANDAG, but given that such capabilites are completely untethered to any empirically observed data I have reservations about this investment, as it is contrary to my own hopes that ActivitySim be truly ""data driven"". I think we as a community should really get serious about collecting observed TNC fleet behavior.
ODOT: In ODOT's testing we have found that it's very easy to get unrealistic results if no constraint or assumptions on the fleet is set. A critical part of routing is knowing the fleet size. I would advocate that if this moved forward that it would have the ability for the user to set a fleet size by time period, as well as an option for the model to calculate the fleet size needed to meet the demand and return that as an output.
SANDAG: it seems this is related HH and Fleed vehcile tracking in task 1?
MTC: See comment B) on 1 "Vehicle tracking and fleet behavior"
Ben: Is this more of supply model like RSG did in FL?
In the case of shared AVs, there are more system-wide options with TNCs and MaaS providers, hence, properly analyzing such options may also require the introduction of a sub-model for vehicle routing between the ActivitySim demand models and the assignment model.
MTC, SANDAG, and others have built AV/TNC supply side / fleet management / routing modules. These modules are similar to transit supply models in that they attempt to model a vehicle fleet to meet travel demand, but build tours to schedule and dispatch service as opposed to using fixed routes. There is not a lot of observed data to estimate these models, in part because TNC operators are secret about their operating plans. TransModeler (and possibly other software providers) have developed related functionality. These features are useful for answering questions related to TNC fleet size, number of drivers, mode share shifts, etc.
Additional Description
Added from
ActivitySim Budget Phase 9.xslx