| --- |
| language: |
| - en |
| --- |
| |
| # Object-Centric Trajectory Dataset (Franka End-Effector in Object Frame) |
| This repository contains a dataset of object-centric cartesian trajectories generated with for Franka robotic arm. Each trajectory represents the motion of the FR3 end-effector expressed in the object coordinate frame. |
| If you want to quickliy visualize outcome archive exemples, please go the following hugging face space : [trajectory_primitive_visualisation_space](https://huggingface.co/spaces/mathildekappel/trajectory_primitive_visualisation_space ) |
|
|
| ## Dataset Overview |
| The dataset of trajectories is organized into outcome archive. Each outcome archive sotres diverse trajectory allowing to activate a specific activation primitive Here is the following folder convention for outcome archive storage: |
| IdObject_TypeJoint_IndexJoint |
|
|
| ## What is an Object-Centric Trajectory? |
| Each trajectory describes the motion of the Franka end-effector relative to an object frame, rather than a world frame. |
|
|
| This means: |
|
|
| - The object is the reference frame |
| - The trajectory encodes how the robot interacts with the object |
| - It is well-suited for generalization across object placements |
|
|
| ## Data Format |
| Each line in the dataset corresponds to a trajectory stored in JSON format. A trajectory contains a sequence of keyframes, and each entry includes the following metadata: |
| Available Fields |
| - object_to_grasp: target object for manipulation |
| - joint_nature: type of joint interaction |
| - item_joint_to_activate: active joint or control target |
|
|
| ## Visualization |
| ### RViz (ROS2 Deployment on Real Robot) |
| Trajectories can be directly visualized in RViz using ROS2. |
|
|
| ### Genesis Replay (QD-Trajectory Viewer) |
| Sotre the outcome archive in other_pkgs/resutls/ and run the replay mode. |
| |
| ### contact |
| kappel@isir.upmc.fr |
