Open source framework for multimodal data collection on a simulated Baxter robot
Prerequisites: Follow the steps for the Baxter Simulator setup guide at http://sdk.rethinkrobotics.com/wiki/Simulator_Installation (this requires Ubuntu 14 and ROS Indigo)
Installation: Clone this repository into ROS_DIRECTORY/src/drumming/ and run catkin_make from the ROS_DIRECTORY.
To launch the simulator with the drum setup, use: roslaunch drumming baxter_drums.launch
To run the collision detection node in a separate process, use: rosrun drumming collision_detection_node.py
To run the data collection framework with default parameters, use: rosrun drumming trajectory_mmd_recorder.py
Data are automatically saved to the trials/ directory as a .mmd file for each trial. This format can be read using the load_mmd_obj() function from the baxter_mmd module, which loads it as an object of class MultimodalData, containing image, audio, and joint attributes, as well as several visualisation methods.