I am using Dymola 6.0b with Modelica 2.2.1 and wondering if it is possible to switch between models during simulation.
The detailed problem would be the following:

I am working on simulating a walking biped robot using components from the Modelica.Mechanics.MultiBody and Modelica.Mechanics.Rotational libraries, at this time WITHOUT collision detection and collision response.
I have attached a GIF screenshot about the 21 DOF humanoid model.
I am using a PointToPoint controller approach with TimeTables to drive the individual ActuatedRevolute joints along a desired path.
DC Servo motors with angle target are also modeled:

In my case a simple stepping motion contains the following phases (substeps):
- double support #1 with both feet side-by-side parallel on the floor ("Fixed Translation" on both feet)
- single support #1: left foot is fixed and the right leg is swinging ("Fixed Translation" on left foot)
- double support #2: both feet on the floor, right foot is ahead by "step length"
- single support #2: right foot is fixed and the left leg is swinging and closing near the right one
- double support #3: both feet are parallel on the floor, side-by-side

I don't think that in this case it was a good approach to maintain only one extra complex "sophisticated" model that does alone all of these 5 supsteps to make the robot do a single walking step.

I think all these substeps require a separate model (with various number of FixedTranslation components).
Would it be possible to concatenate these models during the simulation?
I mean that the result forces / momentums / poses should be somehow the initials for the next model (=next phase)...

If this is not possible then I should consider to incorporate some simple collision response in the big, sophisticated model.