Human-Robot Physical Contact Classification Dataset

Citation Author(s):
Submitted by:
Naixin Kang
Last updated:
Tue, 12/19/2023 - 22:02
0 ratings - Please login to submit your rating.


this is a dataset for Human-Robot Physical Contact Classification. We used the UR10e six-axis robotic arm as the data collection object and the official tool, RTDE, as the data acquisition tool. Regarding the labels of the dataset, we categorize Human-Robot physical contact into three types: no contact, intentional contact, and collision, based on common occurrences in Human-Robot collaborative tasks. The dataset contains 2375 non-repetitive data entries with valid Human-Robot physical contact information, and each entry includes the motion data of the robotic arm within 1 second. Of these, 1800 entries are used for training, and 575 entries are used for validation. Within the entire dataset, there are 800 entries each for Non-Contact and Intentional, as well as 775 entries for Collision.


Just rehsape each piece of data into a 24*28 matrix and feed it directly into the neural network.

Dataset Files

    Files have not been uploaded for this dataset