Sample image MODELING DIVERSE INTERACTIONS IN MULTIMODAL CORRIDORS

Sample image MODELING DIVERSE INTERACTIONS IN MULTIMODAL CORRIDORS Demystifying agent-based modeling as a potential approach

Complexity abound… • Complexity ≠ complicated – Complicated problems have little uncertainty (e. g. , engineered problems such as sending a rocket to the moon or building a bridge, a surgeon performing a procedure) – Complex problems have greater uncertainty and are founded on dynamic interactions among its component parts (e. g. , raising a child, managing traffic congestion) • Unlike simple problems of few variables (e. g. , current, resistance, and voltage), or problems of disorganized complexity (e. g. , understanding laws of temperature and pressure as emerging from trillions of disorganized air molecules) , problems of organized complexity cannot be solved through science of averages or statistical mechanics – Systems of organized complexity involve a moderate number of variables, 2 all interrelated with strong, nonlinear interactions among them

Complexity within systems • Complex adaptive systems (CASs) are characterized by – – Simple components or agents (simple relative to whole system) Nonlinear interactions among components No central control Emergent phenomena due to interactions of the components, based on o Hierarchical organization o Information, communication, and its processing into action o Dynamics of structure and behavior of the system o Evolution and learning within constantly changing environments • Complexity science is the study of how CAS behaviors and outcomes are influenced by these properties 3

Ants as a natural example of complexity • Ants are behaviorally simple as individuals, though cooperatively as a colony, may accomplish complex tasks with no central control 4

The brain as a complex adaptive system • The human brain consists of 100 billion neurons and over 100 trillion connections between them • Neurons are the individual agents with relatively simple behaviors, yet without centralized control, can selforganize to give rise to complex emergent behaviors – – Cognition Intelligence Creativity Personality 5

Cities as a complex adaptive system • Cities resemble living organisms, sensing and adapting to its environment (e. g. , economy, population, politics), and reacting in ways that influence its organizational structure, growth, and dynamics of operation (e. g. , congestion, land-use) • The component agents are its citizens, 6

Data, data, everywhere … • A common approach for assessing impacts in complex systems, is to use a purely empirical basis, using economic models or big data and predictive analytics • Data analytics is useful – Determine key driving factors to system performance – Find variable relationships – Forecast by extending data relationships into the future • However, insights beyond bounds of the data set are needed for more robustly informed decisions 7

Agent-based modeling as a promising approach • Agent-based models are a promising approach to – Better understand complexity of real-world systems – Explore ways to interact with them (e. g. , policy-making) • Agent-based models simulate actors in a system as software agents that interact with their environment and other agents • Agents are – Discrete and heterogeneous entities with their own goals, interdependency relationships, and behaviors – Autonomous with capability to adapt, learn, and modify its behaviors based on some stimulus • Agent-based models have a wide spectrum of application between theoretical and detailed representations of real-world systems 8

Phantom traffic jams • Why are there traffic jams when there is no apparent cause? 9

Traffic jam phenomena and its propagation 10

Agent-based implementation of traffic jams • Simple agent rules: – Accelerate if all clear ahead – Slow down if a car ahead • Traffic jams emerge without any accidents or other artifacts of a “centralized cause” 11

Current implementations in transportation Agent-based Travel Demand Agent Characteristics Demographics Household income Other socioeconomic factors Agent Behaviors Origin / destination Route choice Transit frequency Microsimulation for System Impacts 12

Multimodal corridors • Automobility is not the only mode choice for urban travel • Corridor scope makes interactions important • Challenges – Accounting for complementary multimodalism o Access may limit availability and use of specific modes – Accommodation for public and private modes o Curbside sharing may cause safety conflicts – Asymmetry of a multi-modal commute o Network imbalances, that dynamically Source: http: //nacto. org/publication/urban-street-design-guide/streets/transit-corridor/ 13

Complexity characteristics of multimodal corridors • Simple components or agents – – – Auto Rail Bus Pedestrian Cyclists Increasingly autonomous vehicles • Simple behaviors – Transit objective – Obstacle avoidance and speed change – System rules Source: http: //nacto. org/publication/urban-street-design-guide/streets/transit-corridor/ 14

Complexity characteristics of multimodal corridors • Nonlinear interactions among components – Disproportionate response to proportionally small changes – E. g. , small delays due to some modal interaction (slower auto speeds due to increased bike lane use) may cascade into significant systemic delay Source: http: //nacto. org/publication/urban-street-design-guide/streets/transit-corridor/ 15

Complexity characteristics of multimodal corridors • No central control – No controlling entity or process that dictates how the emergent properties of the system should be formed Source: http: //nacto. org/publication/urban-street-design-guide/streets/transit-corridor/ 16

Complexity characteristics of multimodal corridors • Emergent phenomena due to interactions of the components, for example, system impacts due to: – How pedestrians, auto, rail, and bicycles interact with each other – Mode choice influenced by prior experience – Influences due to social network biases and perspectives of individual connections Source: http: //nacto. org/publication/urban-street-design-guide/streets/transit-corridor/ 17

Potential research approach complexity context • Establish a baseline scenario case with available empirical data useful for validation • Develop an agent-based model for the study area and scope • Validate against baseline to establish control • Experiment – Conduct sensitivity analyses to determine the degree of influence agents and their individual behaviors have on systemic impacts – Explore policy changes, corridor changes (e. g. , access management treatments), land-use changes, population change, etc. • Derive recommendations based on experiment insights 18

Contact Brant Horio LMI bhorio@lmi. org (571) 633 -7838 41