Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Characterizing Structured versus Unstructured Environments based on Pedestrians' and Vehicles' Motion Trajectories

Mahsa Golchoubian, Moojan Ghafurian, Nasser Lashgarian Azad, Kerstin Dautenhahn

TL;DR

The paper addresses the need to classify environments as structured or unstructured using trajectory data from pedestrians and vehicles. It proposes a quantitative framework by extracting 13 motion features across pedestrian, vehicle, and interaction categories, then applying $k$-means clustering to label datasets into two environment types and using generalized linear models to test feature effects. Key findings show significant differences in pedestrian stop fraction, trajectory variability, path efficiency, standing density, vehicle stop fraction, and interaction entropy between clusters, with cluster A aligning with unstructured campus-like environments and cluster B with road-like structured environments. This quantitative environment labeling can guide the selection of training data for autonomous-vehicle trajectory prediction, improving generalization by matching predictors to the target environment type.

Abstract

Trajectory behaviours of pedestrians and vehicles operating close to each other can be different in unstructured compared to structured environments. These differences in the motion behaviour are valuable to be considered in the trajectory prediction algorithm of an autonomous vehicle. However, the available datasets on pedestrians' and vehicles' trajectories that are commonly used as benchmarks for trajectory prediction have not been classified based on the nature of their environment. On the other hand, the definitions provided for unstructured and structured environments are rather qualitative and hard to be used for justifying the type of a given environment. In this paper, we have compared different existing datasets based on a couple of extracted trajectory features, such as mean speed and trajectory variability. Through K-means clustering and generalized linear models, we propose more quantitative measures for distinguishing the two different types of environments. Our results show that features such as trajectory variability, stop fraction and density of pedestrians are different among the two environmental types and can be used to classify the existing datasets.

Characterizing Structured versus Unstructured Environments based on Pedestrians' and Vehicles' Motion Trajectories

TL;DR

The paper addresses the need to classify environments as structured or unstructured using trajectory data from pedestrians and vehicles. It proposes a quantitative framework by extracting 13 motion features across pedestrian, vehicle, and interaction categories, then applying -means clustering to label datasets into two environment types and using generalized linear models to test feature effects. Key findings show significant differences in pedestrian stop fraction, trajectory variability, path efficiency, standing density, vehicle stop fraction, and interaction entropy between clusters, with cluster A aligning with unstructured campus-like environments and cluster B with road-like structured environments. This quantitative environment labeling can guide the selection of training data for autonomous-vehicle trajectory prediction, improving generalization by matching predictors to the target environment type.

Abstract

Trajectory behaviours of pedestrians and vehicles operating close to each other can be different in unstructured compared to structured environments. These differences in the motion behaviour are valuable to be considered in the trajectory prediction algorithm of an autonomous vehicle. However, the available datasets on pedestrians' and vehicles' trajectories that are commonly used as benchmarks for trajectory prediction have not been classified based on the nature of their environment. On the other hand, the definitions provided for unstructured and structured environments are rather qualitative and hard to be used for justifying the type of a given environment. In this paper, we have compared different existing datasets based on a couple of extracted trajectory features, such as mean speed and trajectory variability. Through K-means clustering and generalized linear models, we propose more quantitative measures for distinguishing the two different types of environments. Our results show that features such as trajectory variability, stop fraction and density of pedestrians are different among the two environmental types and can be used to classify the existing datasets.

Paper Structure

This paper contains 9 sections, 1 equation, 9 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Datasets
  4. Method
  5. Extracted trajectory features
  6. Data Analysis Approach
  7. Results and Discussion
  8. Limitations and Future Work
  9. Conclusion

Figures (9)

  • Figure 1: Approach angle definition in a pedestrian-vehicle interaction
  • Figure 2: The value of features extracted from pedestrians' trajectories in each dataset with 95 % confidence interval
  • Figure 3: The value of features extracted from vehicles' trajectories in each dataset with 95 % confidence interval
  • Figure 4: The value of features extracted from pedestrian-vehicle interaction in each data
  • Figure 5: Datasets clustered in two groups based on the defined features. For each dataset the portion of its data points in each cluster is reported.
  • ...and 4 more figures