Detection of interaction forces in industrial robotics
Детекција сила интеракције у индустријској роботици
Докторанд
Gordić, ZavišaМентор
Jovanović, KostaЧланови комисије
Đurović, ŽeljkoRodić, Aleksandar
Miljković, Zoran
Jovičić, Nenad
Метаподаци
Приказ свих података о дисертацијиСажетак
With Industry 4.0 becoming a reality and Industry 5.0 emerging on the horizon, the
need for seamless integration, shared workspace and interoperability of production entities is ever
increasing. To aid in this transition, this thesis presents approaches intended to allow the evolution of
industrial robots by enabling them to detect and interpret interactions with their surroundings. The
detection of interaction forces is based on non-model-based algorithms due to their inherent ability to
include all aspects of the behaviours of the robot as well as to capture the contact task-specific forces
and dynamics. To detect interactions, the reference sequence recorded during an exemplary task
execution cycle is compared with measurements from the robot while it is performing its repetitive task.
The thesis presents several different approaches to detection of collisions and interactions in general
intended for the implementation on industrial robots with closed control architecture. To overco...me
implementation issues, the modified Dynamic Time Warping (mDTW) method, as one of the key
presented contributions, enables optimal matching of compared signals. The mDTW enables comparing
a signal with the most similar section of the other signal. Partial matching also enables online
application of time warping principles and reduces the time and computation resources needed to
perform matching. The developed and presented algorithms for automatic calculation of kinematic
parameters of the robot and its end-effector enable further evolution of the mDTW in into its
kinematically augmented version - KA-mDTW, extending the interaction’s detection algorithm’s
application domain. Furthermore, it enables the inclusion of unmodeled task dynamics or a robot’s endeffector into algorithms for collision detection or general understanding of a robot’s operation context.
The presented algorithms and conclusions are supported and validated by the experimental testing on
industrial robots.