Re: FSM vs. Petri nets for component modelling...

We have such kind of reconfiguration in Orccad, where the whole running 
control law can be stopped and replaced by a new one during task execution. 
Thus control laws are encapsulated in a logical behaviour stating starting, 
stopping and exception handling. As hand writting of FSMs by hand  is 
difficult and error prone, this logical behaviour is specified with Esterel, 
thus most of the job is done by the strl compiler. Obviously this model can 
handle control tasks running in parallel.

See for example: 

K. Kapellos and D. Simon and M. Jourdan and B. Espiau: "Task
level specification and formal verification of robotics control
systems: state of the art and case study", Int. Journal of Systems
Science vol. 30, no 11, pp 1227--1245, november 1999.