학술논문

NZFIK: A Fast Method for Human Pose Estimation
Document Type
Conference
Source
2023 International Conference on Artificial Intelligence and Automation Control (AIAC) AIAC Artificial Intelligence and Automation Control (AIAC), 2023 International Conference on. :133-138 Nov, 2023
Subject
Computing and Processing
Wrist
Automation
Pose estimation
Kinematics
Approximation algorithms
Distortion
Iterative algorithms
Inverse kinematics
Human pose estimation
Endeffector
Motion capture
Language
Abstract
This article proposes a non-iterative zero-error fast inverse kinematics algorithm (NZFIK) based on the follow the leader (FTL) algorithm for human motion restoration with a small amount of sensor data. It can calculate joint positions point by point with the strategy of minimizing root pose changes. The algorithm can quickly calculate typical joint connection forms such as chain and closed-loop, and compared with the forward and backward reaching inverse kinematics (FABRIK) algorithm, it significantly improves computational efficiency while obtaining approximate results. Finally, this article takes the rigid connection solution of human upper limb joints as an example to validate the effectiveness of the algorithm. The results show that the NZFIK algorithm has good performance in restoring human upper limb posture based on a small number of end effectors on the wrist and head.