Dynamic Simulation for Robot Grasping and Manipulation
Dynamic simulation of grasping and manipulation can be very helpful in the development of algorithms and hardware. The Mechatronics group has experience from working in GraspIt! and Adams. This page explains some of the basics and show some demonstrations.
The Purpose of Simulation
The interplay between robotic hardware, control algorithms and the surroundings is complex. In this and other situations, simulation can be a very helpful tool. Simulations can help us understand what is going on, they allow us to test new ideas without all the complivations associated with hardware and reality. In addition – if the simulations correspond well to reality – they can be used to generate experience that can be used for learning and system specification.
For most of our work, we have been using an extended version of GraspIt!. By adding tactile sensor models and filtering contact information from GraspIt!, it is possible to simulate tactile sensors. Thanks to the socket server of GraspIt!, is also possible to perform control and interact with GraspIt! from another software. We choose to implement all control in Matlab and/or Simulink.
The KTHand – our preferred robot manipulator – is underactuated. As GraspIt! has no support for that, also the underactuation has been modelled in Matlab.
A student project has also investigated dynamic simulation in ADAMS. Please check out the videos below!
Feel free to ask for our reports or thesises if you want more detail.