Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Data-driven Exploration of Mobility Interaction Patterns

Gabriele Galatolo, Mirco Nanni

TL;DR

This paper introduces a data-driven framework (IPA) to study mobility interactions by extracting interaction events from trajectory data. It formalizes a neighbor-graph based representation, defines event templates and both static and evolving interaction patterns, and provides SIPM and EvIPM algorithms to mine frequent patterns. The methods are instantiated on two real datasets (NGSIM vehicles and campus pedestrians), yielding interpretable patterns such as overtaking, following, and coordinated groups, and showing parameter sensitivity and scalability. The work offers a data-first alternative to model-driven crowd and traffic simulations with potential to refine existing models and improve emergency management planning.

Abstract

Understanding the movement behaviours of individuals and the way they react to the external world is a key component of any problem that involves the modelling of human dynamics at a physical level. In particular, it is crucial to capture the influence that the presence of an individual can have on the others. Important examples of applications include crowd simulation and emergency management, where the simulation of the mass of people passes through the simulation of the individuals, taking into consideration the others as part of the general context. While existing solutions basically start from some preconceived behavioural model, in this work we propose an approach that starts directly from the data, adopting a data mining perspective. Our method searches the mobility events in the data that might be possible evidences of mutual interactions between individuals, and on top of them looks for complex, persistent patterns and time evolving configurations of events. The study of these patterns can provide new insights on the mechanics of mobility interactions between individuals, which can potentially help in improving existing simulation models. We instantiate the general methodology on two real case studies, one on cars and one on pedestrians, and a full experimental evaluation is performed, both in terms of performances, parameter sensitivity and interpretation of sample results.

Data-driven Exploration of Mobility Interaction Patterns

TL;DR

This paper introduces a data-driven framework (IPA) to study mobility interactions by extracting interaction events from trajectory data. It formalizes a neighbor-graph based representation, defines event templates and both static and evolving interaction patterns, and provides SIPM and EvIPM algorithms to mine frequent patterns. The methods are instantiated on two real datasets (NGSIM vehicles and campus pedestrians), yielding interpretable patterns such as overtaking, following, and coordinated groups, and showing parameter sensitivity and scalability. The work offers a data-first alternative to model-driven crowd and traffic simulations with potential to refine existing models and improve emergency management planning.

Abstract

Understanding the movement behaviours of individuals and the way they react to the external world is a key component of any problem that involves the modelling of human dynamics at a physical level. In particular, it is crucial to capture the influence that the presence of an individual can have on the others. Important examples of applications include crowd simulation and emergency management, where the simulation of the mass of people passes through the simulation of the individuals, taking into consideration the others as part of the general context. While existing solutions basically start from some preconceived behavioural model, in this work we propose an approach that starts directly from the data, adopting a data mining perspective. Our method searches the mobility events in the data that might be possible evidences of mutual interactions between individuals, and on top of them looks for complex, persistent patterns and time evolving configurations of events. The study of these patterns can provide new insights on the mechanics of mobility interactions between individuals, which can potentially help in improving existing simulation models. We instantiate the general methodology on two real case studies, one on cars and one on pedestrians, and a full experimental evaluation is performed, both in terms of performances, parameter sensitivity and interpretation of sample results.

Paper Structure

This paper contains 14 sections, 2 theorems, 2 equations, 11 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Interaction Pattern Analysis
  3. Basic Framework and Interaction Events
  4. Static Interaction Patterns
  5. Evolving Patterns
  6. Pattern Mining Algorithms
  7. Computing Frequent Static Patterns
  8. The EvIPM algorithm
  9. Experiments
  10. Analysis on NGSIM vehicle dataset
  11. Analysis on a campus crowd dataset
  12. SIPM and EvIPM performances
  13. Related works
  14. Conclusion

Key Result

Theorem 1

Given a set of event instances $\mathcal{EI}$ observed in a time scope T and a minimum persistence time for each instance pattern of $t_{min}$, the time complexity of the SIPM algorithm is $O(m \cdot (4n)^n \cdot n^3)$, where $n$ is the maximum number of different agents present in one instant and

Figures (11)

  • Figure 1: Left: instances of a complex event involving different agents. Right: pattern representing the two instances.
  • Figure 2: A complete overtaking: an approach(A,B) followed by a flanking(A,B) and finally by a moving_away(A,B)
  • Figure 3: Static Interaction Pattern Mining pseudo-code
  • Figure 4: Candidate Generation pseudo-code
  • Figure 5: Evolving Interaction Pattern Mining pseudo-code
  • ...and 6 more figures

Theorems & Definitions (18)

  • Definition 1: Agent set and agent
  • Definition 2: Neighbor Graph
  • Definition 3: Ego and interaction functions
  • Definition 4: Labelled Graph
  • Definition 5: Event template
  • Example 1
  • Definition 6: Quasi-continuous events
  • Definition 7: Interaction Events
  • Definition 8: Static Pattern Instance
  • Definition 9: Isomorphic Instances
  • ...and 8 more