Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 41


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 40
28 I N N O VAT I O N S I N T R AV E L D E M A N D M O D E L I N G , V O L U M E 1 The process results in an update of an agent's needs and across the sequence of travel and activity episodes that it gaining knowledge from the experience. contains. Utility is dependent on the time of the day, the A prototype activity-based model of transport activity duration, when the activity was performed, and demand for Flanders, Belgium, called Feathers, will the time since the previous activity. extend the Aurora model and add complementary con- The input of the scheduling heuristic is a consistent cepts. The project is part of a wider research program schedule in terms of duration and timing choices. The involving a number of research institutes. Other elements output is also a consistent schedule with utility that is being examined include the application of combined higher or equal to that of the inputs. The model itera- Global Positioning System (GPS) and personal digital tively implements operations on an existing schedule assistant (PDA) technology for collecting activity-travel until no further improvement is possible. Operations are data (called PARROTS for PDA system for an Activity evaluated under optimal duration and timing choices. Registration and Recording of Travel Scheduling) and Operations considered include inserting an activity, sub- the use of new technology to collect vehicle data. stituting an activity, and repositioning an activity. Other Additional contributions to Feathers expected as possible operations are changing the location of an activ- part of the ongoing research program include modeling ity, including or removing a return-home trip between route choice behavior through the data obtained from activities, and changing the mode of a trip. PARROTS and calibrating the current model based on Uncertainty is dependent on an agent's attitude real-world data. Research will also test and improve with respect to risk. Various decision-making principles currently used concepts of Aurora, such as estimating S- can be accommodated within the model. Agents hold curves as utility functions, estimating the effect of con- beliefs or subjective probabilities with respect to the text variables on maximum utility, evaluating the expected state of system variables. Beliefs are represented scheduling component, and extending learning facets. by a probability distribution across possible system Additional concepts are also anticipated to be imple- states. The expected utility of a schedule alternative is mented in Feathers, including the impact of life trajec- the weighted sum of the utilities of the schedule, depen- tory events, which include events such as getting dent on the state variables, where the weights represent married and starting a job. Finally, research elements the beliefs. will focus on guiding and helping practitioners with the There are different types of learning. Attribute transition from four-step models to activity-based learning is the simplest form of learning. Agents learn models. about their environment based on their expectations. Aurora is an agent-based microsimulation system Agents update their beliefs about states of single system in which each individual in the population is represented variables. Conditional learning relates to updating causal as an agent. It is an activity-based model that simulates knowledge. For example, differences in travel time can the full pattern of activity and travel episodes of each be explained by the day of the week and the time of agent for each day of the simulated time period. The travel. Associative learning results from generalization. dynamics of the Aurora system start at the beginning of It means an agent's beliefs can change or remain the same the day for each agent. The schedule is implemented based on experience. Information-based learning is based on the needs and knowledge of each agent. The based on information sources such as the news media environment has an impact on the implementation of the and agency announcements. The impact of this informa- schedule for each agent in time and space. There is inter- tion depends on the credibility that agents place in the action between agents who are competing against each source. Social learning means that agents learn from other, which is when congestion occurs. Some agents members in their social network. decide to reschedule their original schedule based on Issues that have been identified to date relate to the their needs and knowledge. synthetic population, belief updating, and other ele- The scheduling and rescheduling model assumes ments. The system shows how an activity-based model that activities and travel are scheduled on a continuous can be used for microsimulation of spatial behavior. The time scale. The schedule meets a full set of scheduling framework embraces and integrates the urgency of activ- constraints for each agent. Needs for activities grow over ities as a function of time, time budgets and competition time and are satisfied by activities depending on dura- between activities, spacetime constraints, the ability to tion. Scheduling decisions are based on heuristics, rather reschedule activities, the ability to learn from interaction than on an exhaustive search. Inputs to the scheduling with the environment, and the ability to deal with uncer- model include utility functions, dynamic constraints, tainty. The system allows users to analyze impacts of activity needs, and knowledge of the land use and trans- temporal as well as spatial variables on utilities and traf- portation systems. fic flows. The model is based on a set of utility functions. Other aspects will be added to the system within The utility of a schedule is defined as the sum of utilities the context of Feathers. These aspects include explicitly