humanObjectDetection
收藏资源简介:
本数据集用于训练和验证机器学习模型,以区分机器人与三种不同类型物体(人类、铝材和PVC)的接触。数据由Franka Emika Panda机械臂在实时交互中收集,采样率为200Hz。数据集包括85个时间序列用于训练和40个时间序列用于验证。每个时间序列记录了机器人与各自物体之间的多次接触,每个序列添加了3次接触。数据集的创建旨在帮助机器人更好地理解和适应其工作环境,从而提高人机协作的安全性。
This dataset is used to train and validate machine learning models for distinguishing contacts between a robot and three distinct types of objects: humans, aluminum materials, and PVC. The data was collected in real-time interactions using a Franka Emika Panda robotic arm, with a sampling rate of 200 Hz. The dataset comprises 85 time series for training and 40 time series for validation. Each time series records multiple contacts between the robot and its corresponding object, with 3 contact instances added per sequence. This dataset was developed to help robots better understand and adapt to their working environments, thereby improving the safety of human-robot collaboration.
数据集概述
数据集名称
Contact Interpretation System
依赖库
- numpy==1.19.5
- torch==1.10.1
- pandas==1.1.5
- torchvision==0.11.2
- torchmetrics==0.8.2
- canlib
机器人API
Franka Emika Panda
- franka-interface: Welcome to franka-interface’s Documentation! — franka-interface 1.0.0 documentation (iamlab-cmu.github.io)
- frankapy: Welcome to FrankaPy’s Documentation! — frankapy 1.0.0 documentation (iamlab-cmu.github.io)
Universal Robots
- Universal Robots RTDE: https://sdurobotics.gitlab.io/ur_rtde/
运行机器人和接触检测
Franka Robot
-
解锁机器人
- 打开机器人(等待直到显示稳定的黄色)
- 连接到机器人桌面(ID:172.16.0.2或192.168.15.33)
- 解锁机器人(灯光应为蓝色)
- 解锁机器人并激活FCI
-
运行frankapy
- 打开终端
- 激活conda环境:
conda activate frankapyenv - 运行脚本:
bash /home/mindlab/franka/frankapy/bash_scripts/start_control_pc.sh -i localhost
-
运行数字手套节点
- 打开另一个终端
- 设置ROS环境:
source /opt/ros/noetic/setup.bash - 运行脚本:
/home/mindlab/miniconda3/envs/frankapyenv/bin/python /home/mindlab/humanObjectDetection/dataLabeling/sensorNode.py
-
运行机器人节点
- 打开另一个终端
- 激活conda环境:
conda activate frankapyenv - 设置ROS环境:
source /opt/ros/noetic/setup.bash - 扩展环境:
source /home/mindlab/franka/franka-interface/catkin_ws/devel/setup.bash --extend - 扩展环境:
source /home/mindlab/franka/frankapy/catkin_ws/devel/setup.bash --extend - 运行脚本:
/home/mindlab/miniconda3/envs/frankapyenv/bin/python3 /home/mindlab/humanObjectDetection/frankaRobot/main.py
-
运行保存数据节点
- 打开另一个终端
- 设置ROS环境:
source /opt/ros/noetic/setup.bash - 扩展环境:
source /home/mindlab/franka/franka-interface/catkin_ws/devel/setup.bash --extend - 扩展环境:
source /home/mindlab/franka/frankapy/catkin_ws/devel/setup.bash --extend - 运行脚本:
/home/mindlab/miniconda3/envs/frankapyenv/bin/python3 /home/mindlab/humanObjectDetection/frankaRobot/saveDataNode.py
修改frankastate发布频率
- 编辑文件:
sudo nano /home/mindlab/franka/franka-interface/catkin_ws/src/franka_ros_interface/launch/franka_ros_interface.launch
UR Robot
-
激活远程控制
- 从机器人教学面板激活远程控制
-
运行程序
- 打开终端
- 激活conda环境:
conda activate frankapyenv - 设置ROS环境:
source /opt/ros/noetic/setup.bash - 运行脚本:
/home/mindlab/miniconda3/envs/frankapyenv/bin/python3 /home/mindlab/humanObjectDetection/urRobot/main_ur10.py
记录Franka机器人的新动作
-
解锁机器人
- 同Franka Robot步骤1
-
运行frankapy
- 同Franka Robot步骤2
-
运行记录新动作脚本
- 打开另一个终端
- 激活conda环境:
conda activate frankapyenv - 设置ROS环境:
source /opt/ros/noetic/setup.bash - 扩展环境:
source /home/mindlab/franka/franka-interface/catkin_ws/devel/setup.bash --extend - 扩展环境:
source /home/mindlab/franka/frankapy/catkin_ws/devel/setup.bash --extend - 运行脚本:
/home/mindlab/miniconda3/envs/frankapyenv/bin/python3 /home/mindlab/humanObjectDetection/frankaRobot/recordNewMotion.py

- 1Multi-Class Human/Object Detection on Robot Manipulators using Proprioceptive Sensing瑞士苏黎世大学信息学院和ZHAW机械系统研究所MINDLab · 2025年



