From Risk-Taking Tendencies to Congestion and Collision Formation.
Open Access DepositedCognitive Based Modeling of Mixed Heterogeneous Roadway Traffic
Downloadable Content
(5) smooth multilane change trajectories emerge when polynomial lateral curves guide lane tran- sitions
and (6) mixed traffic interactions reveal throughput shifts as bicycle and pedestrians penetration rates exceed critical thresholds with permitted movements. The modeling, calibration/validation and simulation tasks performed in this dissertation are translated to the following contributions
(ii) Utilizing a suitable calibration methodology and making sure that the suggested model reproduces rear-world trajectories especially during differ- ent traffic conditions across roadway users’ groups
(2) bidirectional pedestrian flows self-organize into variable lane clusters without explicit lane assignment
(1) micromobility agents negotiating bottlenecks attain flow capacities and jam densities consistent with empirical benchmarks
(4) vehicular shockwaves prop- agate at speeds closely matching field observations
and (iii) Investigating the macroscopic relationship of the presented calibrated model with differ- ent values of collision weight. With such contributions, this formulation framework empowers planners and engineers to evaluate the safety and efficiency trade-offs of different infrastructure designs, including dedicated cycle lanes, pedestrian refuges, and dynamic speed controls. Despite its advances, the framework has limitations. Calibration datasets are drawn from a limited set of urban environments
allowing agents to adapt dynamically under stress or congestion could enhance the model performance. Group -level behav- iors, such as coordinated pedestrian movements and vehicular platooning, present further opportunities for extension. Lastly, scaling the framework for real- time, city-wide simulations will require algorithmic optimizations and parallel computing implementations.
micromobility trajectories from Amsterdam’s Central Station in Amsterdam, the Netherlands, and mixed trajectories from a freeway and an arterial set-up in Washington, DC, USA. A genetic algorithm is employed to identify parametric values that minimize discrepancies between simulated and observed inter agent spacing. Calibration results demonstrate that the framework achieves relative errors below nine percent for micromobility users and below sixteen percent for motorized vehicles. Cross validation across different sites confirms that the calibrated parameters generalize across contexts, and analysis of parameter interdependencies justifies treating collision weight parameters indepen- dently for each mode. The calibration exercise is followed by a series of simulation experiments that highlight the model’s ability to reproduce key traffic phenomena
(i) a unified cognitive framework that eliminates ad- hoc collision heuristics
Due to the inherent safety concerns associated with traffic movement inunconstrained two- dimensional settings, pedestrians’, bicyclists’, motorized vehicles’ and other modes’ movements are necessary to be modeled as a risk-taking stochastic dynamic process that may lead to errors and thus, contacts and collisions. Among the existing traffic models that may capture risk taking behaviors are
additional contexts such as high- density non-lane-based city centers or rural arterials may exhibit distinct risk perceptions. The model currently treats risk evaluation parameters as static per agent type
(3) evacuation lead to jams that take the form of half-circles near the evacuation exit
2) and the discrete-choice models (through the rationality or the bounded rationality paradigm while weighing different alternatives). Given that the social force models may not readily capture the contact/collision dynamics through the Newtonian force framework, decision-making theories are considered as a feasible approach to formulate a new model that can account for cognitive and behavioral dimensions such as uncertainty and risk. However, instead of relying on the bounded rationality theory, this dissertation introduces a unified microsimu- lation framework grounded in Prospect Theory to model heterogeneous road -users’ behaviors in unconstrained two-dimensional traffic environments. By embedding each agent’s risk-taking attitudes directly into the decision process, the framework allows both collisions and near-misses to emerge naturally from the same subjective valuation mechanism that governs op- erational movement, rather than relying on external, heuristic collision avoidance rules. The core model is based on a utility function for each agent that balances desired motion objectives (such as speed and direction) against the perceived probability and severity of potential collisions. Prospect Theory informs how agents evaluate gains and losses in their movement choices, capturing empirically observed risk aversion and loss aversion behaviors. A separate collision weight parameter quantifies the seriousness of potential impacts, ensuring that risk considerations appropriately influence route deviations and speed adjustments. The model is calibrated and validated while leveraging three empirical datasets
micromobility trajectories from the Lincoln Memorial in Washington, DC, USA
1) the social force models (through the interplay of the repulsion and the attraction force parameters)
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.