In many cases, simulating a dynamical system is best achieved by defining several modes in which it can exist and modeling each mode independently. This approach is known as "hybrid modeling". There is significant academic literature on the subject, but at its most basic level, a hybrid model needs three pieces of information:
- Dynamic models for each domain.
- Rules for when to transition out of each domain.
- A transition graph dictating which domain the system should enter on leaving the previous domain.
Matlab supplies some basic low-level tools for evaluating hybrid systems of this form, and there are heavy-duty toolboxes such as http://www.dii.unisi.it/hybrid/toolbox/. The code provided here falls between these two levels and acts as a lightweight, high-level abstraction for simple-to-medium complexity hybrid systems.
This release also includes two demonstrations of hybrid_integrator.m applied to a bouncing ball: one modeled with a coefficient of restitution at impact, and a second with a spring-damper contact model. |
