A Probabilistic Steering Parameter Model for Deterministic Motion Planning Algorithms
Loading...
Date
2019
Journal Title
Journal ISSN
Volume Title
Publisher
© 2019 The Eurographics Association and John Wiley & Sons Ltd.
Abstract
The simulation of two‐dimensional human locomotion in a bird's eye perspective is a key technology for various domains to realistically predict walk paths. The generated trajectories, however, are frequently deviating from reality due to the usage of simplifying assumptions. For instance, common deterministic motion planning algorithms predominantly utilize a set of static steering parameters (e.g. maximum acceleration or velocity of the agent) to simulate the walking behaviour of a person. This procedure neglects important influence factors, which have a significant impact on the spatio‐temporal characteristics of the finally resulting motion—such as the operator's physical conditions or the probabilistic nature of the human locomotor system. In overcome this drawback, this paper presents an approach to derive probabilistic motion models from a database of captured human motions. Although being initially designed for industrial purposes, this method can be applied to a wide range of use cases while considering an arbitrary number of dependencies (input) and steering parameters (output). To underline its applicability, a probabilistic steering parameter model is implemented, which models velocity, angular velocity and acceleration as a function of the travel distances, path curvature and height of a respective person. Finally, the technical performance and advantages of this model are demonstrated within an evaluation.The simulation of two‐dimensional human locomotion in a bird's eye perspective is a key technology for various domains to realistically predict walk paths. The generated trajectories, however, are frequently deviating from reality due to the usage of simplifying assumptions. For instance, common deterministic motion planning algorithms predominantly utilize a set of static steering parameters (e.g. maximum acceleration or velocity of the agent) to simulate the walking behaviour of a person. This procedure neglects important influence factors, which have a significant impact on the spatio‐temporal characteristics of the finally resulting motion—such as the operator's physical conditions or the probabilistic nature of the human locomotor system. In overcome this drawback, this paper presents an approach to derive probabilistic motion models from a database of captured human motions. Although being initially designed for industrial purposes, this method can be applied to a wide range of use cases while considering an arbitrary number of dependencies (input) and steering parameters (output).
Description
@article{10.1111:cgf.13591,
journal = {Computer Graphics Forum},
title = {{A Probabilistic Steering Parameter Model for Deterministic Motion Planning Algorithms}},
author = {Agethen, Philipp and Gaisbauer, Felix and Rukzio, Enrico},
year = {2019},
publisher = {© 2019 The Eurographics Association and John Wiley & Sons Ltd.},
ISSN = {1467-8659},
DOI = {10.1111/cgf.13591}
}