Meeting Notes

Notes from today’s meeting with Wyatt and Greg

• Greg has a working (ish) model of the robot in Gazebo based on my URDF and STL files.
  • He has defined movable joints in the arms and the wheels
  • He accomplished this by hacking apart the STL file for the base to allow the wheels to rotate.
  • I should create some new STL files of the base that are actually articulable and have appropriate weights/moments instead of his hacked files.
  • Note: Gazebo requires all links have a mass.
  • Note: Gazebo requires all movable joints in URDF files to have max velocity and max torque.
• We discussed goals and project scope
  • Continue with goal of creating a kitting robot.
  • Attempt to make things as easy as possible for the robot to be used as a proof of concept vehicle for the DARPA challenge.
  • Deliverables: Robot with realistic (within reason) model for further use for DARPA or kitting research.
  • Some kind of proof of concept task (Wyatt suggested recreating the soda-can collection work that was done years ago at MIT(?).)
  • Enable low-bandwidth, high level control e.g. Go to Waypoint, Investigate an Object. A student named Steve (last name forgotten) has been working on voice control for smart wheelchairs. His work could be relevant.
• More on Gazebo
  • Gazebo is an open source 3d simulation tool for robotics.
  • Greg has started playing with it, and it was used for the DARPA ARM challenge.
  • DARPA plans to use it (with modifications and improvements) for the new DARPA Robotics challenge
  • It is fairly processor and graphics intensive. Greg has spec'd out some ~$10k machines for if they get funded for DARPA, but in the mean time we may repurpose the solidworks machine for Gazebo. It has an i7 in it and I can drop in a more powerful graphics card.
  • I should learn gazebo and get my robot working in it.