Modeling, prediction, and anomaly detection of manned-vehicle behavior in open field based on velocity vector and variance tensor fields. (18th March 2020)

• Record Type: Journal Article
• Authors: Yamamoto, Ko; Kuwada, Shinsaku; Nakamura, Yoshihiko
• Is Part Of: Advanced robotics. Volume 34:Number 6(2020)
• Page Start: 343
• Page End: 357
• Publication Date: 2020-03-18
• Subjects: Field robotics -- motion prediciton -- anomaly detection -- vector field -- tensor field; Robotics -- Periodicals; Robotics -- Japan -- Periodicals; Robotics; Japan; Periodicals; 629.89205
• DOI: 10.1080/01691864.2019.1709895
• Languages: English
• ISSNs: 0169-1864

Abstract

ABSTRACT: The introduction of autonomous delivery systems is expected to improve productivity in numerous application fields. Since such systems comprise both unmanned and manned units, it is important to control the whole system using appropriate behavioral models of manned vehicles or humans. This work builds on the results of a previous study, where a velocity vector field was used to model human movement, and extends the use of velocity vector fields to the simulation of vehicle behavior by modeling variance in given trajectories. This extension enables the velocity vector field to regenerate an average trajectory with a band and consequently allows one to calculate the vehicle's predicted region in arbitrary future time steps and anomaly detection. These applications are expected to facilitate collision avoidance and should help one to control other applications of unmanned vehicles. GRAPHICAL ABSTRACT: