Multi Objective Control
The controller framework takes in multiple controllers with their states and phases, Each state is represented by its respected stance. For synchronizing multiple controller, current requirement is that they should share the same structure in terms of states and stances associated with them.
We use Open Dynamics Engine (ODE) as our forward dynamics simulator. All character and control parameters are kept fixed during the simulations.First based on the list of contact points, it is determined when to transition the controller to a new state, It also calculates the position of the Center of Mass(COM) of the stance foot in the world coordinates.
For each update cycle the controller takes in a list of contact points, and based on the criteria defined, which can range from user input to various weights attributed to various controller or explicit definition of the interpolation value for the duration of the motion, the controller are blended and processed.
The base framework is of SIMIBICON, Our method for multiple objective control is implemented on top of it.
Examples consists of four different scenarios with three different styles and Interp Value or Blending values of the range 0.0, 0.3, 0.6 and 0.9
- Scenario 1 : Controller Forward Stylized Walk, Jogging and Walking.
- Scenario 2 : Controller Jogging, Walking and Sad walk.
- Scenario 3 : Controller Jollyjog, Jogging and Sad Walk.
- Scenario 4 : Controller Marching, Walking and Jogging.
External Links :
- SIMBICON: Simple Biped Locomotion Control
- 2015 - Updated Article - Uniform Buffers versus Texture Buffers
Source On GitHub (Coming Soon)