Spatiotemporal models for motion planning in human populated environments

Vintr, Tomas and Molina Mellado, Sergi and Cielniak, Grzegorz and Duckett, Tom and Krajnik, Tomas (2017) Spatiotemporal models for motion planning in human populated environments. In: Student Conference on Planning in Artificial Intelligence and Robotics (PAIR), 17 September 2017, Žilina, Slovakia.

Documents
31052 vintr-et-al-pair-2017.pdf
[img]
[Download]
[img]
Preview
PDF
31052 vintr-et-al-pair-2017.pdf - Whole Document

223kB
Item Type:Conference or Workshop contribution (Paper)
Item Status:Live Archive

Abstract

In this paper we present an effective spatio-temporal model for motion planning computed using a novel representation known as the temporary warp space-hypertime continuum. Such a model is suitable for robots that are expected to be helpful to humans in their natural environments. This method allows to capture natural periodicities of human behavior by adding additional time dimensions. The model created thus represents the temporal structure of the human habits within a given space and can be analyzed using regular analytical methods. We visualize the results on a real-world dataset using heatmaps.

Keywords:Long-term Autonomy, Robotic Mapping, Human-Robot Interaction
Subjects:G Mathematical and Computer Sciences > G700 Artificial Intelligence
Divisions:College of Science > School of Computer Science
Related URLs:
ID Code:31052
Deposited On:13 Feb 2018 13:05

Repository Staff Only: item control page