Data and code for kinematic prior-based residual dynamics modeling of robotic Oolong tea shaking

Published: 13 July 2026| Version 1 | DOI: 10.17632/6md3gggkdn.1
Contributors:
Jun Li, Yongwei Song, Zhiqin Guo, Zhaoyang He, Pingwei Jiang, Zhonghua Liu, Zhou Fang, Yinhui Xie

Description

This dataset supports the manuscript entitled “Kinematic Prior-Based Residual Dynamics Modeling for Phase-Consistent Spatiotemporal Trajectory Prediction in Robotic Oolong Tea Shaking.” The dataset contains raw and processed three-dimensional trajectory data collected using a robotic Oolong tea shaking platform under multiple operating conditions. The experimental platform employs a universal-joint-constrained spatial linkage with independently adjustable effective crank length and rotational speed. The trajectory data were acquired using a three-dimensional motion measurement system and were subsequently processed for geometric phase alignment and trajectory prediction analysis. The deposited materials include original measurement files, standardized trajectory data, phase-aligned trajectory data, geometric parameter identification results, trained model files, evaluation results, and Python source code. The code implements spatial Fourier trajectory reconstruction, L-BFGS-B-based geometric parameter calibration, multilayer perceptron and long short-term memory residual modeling, and trajectory prediction performance evaluation. Supporting documentation is also provided, including README files, a data dictionary, file manifests, metadata templates, software requirements, and instructions for reproducing the main data-processing and modeling procedures. The dataset is intended to support verification, reproduction, and further development of phase-consistent mechanism–data hybrid modeling methods for robotic food-processing equipment.

Files

Categories

Mechanical Engineering, Robotics, Food Engineering, Machine Learning

Licence