unipi-grasp-datasets

This pacakge contains experiment protocols on how to generate grasp datasets. Everything is based on ROS Indigo/Ubunt 14.04. You must have a catkin workspace configured in order to follow the instructions.

Equipment & Dependencies

• PhaseSpace
• Asus/Kinect (install from synaptic/apt-get recommended)
• F/T
• Intrinsic Tactile Toolbox
• SoftHand
• Flexiforce Glove
• Calibration
• PaCManObjectDatabse

You need to add them all to your catkin workspace to compile messages in the sync package.

Guidelines to create grasp experiment scenarios

• If you use additional equipments or dependencies, please, add the link where to find them in the list above
• Create a package of your scenario using catkin_create_pkg <scnarioname> (more arguments can be useful to fill the package.xml)
• Follow the template of scenario1, create at least these files:
  • launch/uploadSetup.launch to bring up all devices
  • data/startRecording.sh to define your recording protocol (if possible, record only sensor data and parameters)
  • launch/playExperiment.launch to play back recorded data
• IMPORTANT: Do not commit your bag files to this repository! One alternative is to upload your bag files to the CentroPiaggio server under the datasets folder as a single zip, share the file and provide a public link. If you don't have an account, ask to the technical dept. or your supervisor.
• EVEN MORE IMPORTANT: Document your scenario in a README file and add it to the list below. Fill at least the following two sections:
  • Experiment protocol to explain how the experiments must be ran/were ran
  • How to play the recorded data to describe how to play the data you obtained and provide download links to your bag files.
• The package sync_data allow you to synchronize your data either for recording (not recommended) or for data being played. To this end, you should:
  • Create msg/<scenarioname>.msg file to define which type of message you want to synchronize
  • Write a src/<scnarioname>sync.cpp node to subscribe to the topics you want to synchronize
  • Create a launch/<scenarioname>.launch file to facilitate the use

Scenarios

PaCMan - WP2 Grasps: Human grasping using a handled Pisa/IIT SoftHand and point cloud scene

This scenario has been designed to fulfil requirements in WP2 of the PaCMan project. Check the README file to see where to download the data, how was recorded, and how can it be played back.

IROS2015 - submitted: Human grasping using a handled Pisa/IIT SoftHand

This scenario has been designed to create the Grasp database necessary for the ``High-Level Planning for Dual Arm Goal-Oriented Tasks'' work submitted to IROS2015. Check the package README file to see where to download the data, how was recorded, and how can it be played back.