MultiBody and RigidBody: architectural design of an application
Posted: Wed Mar 13, 2019 11:23 am
Hello,
This is a general question in order to approach the development of an application using MultiBody and/or RigidBody.
Think to a vehicle and some different articulated actuators (excavator, shovel, etc...). I would like be able to model the vehicle physic and then add the actuators to this one, dynamically (at run-time) . I can think to these approaches:
1. Vehicle is a RigidBody and actuators are a chain of RigidBody (using constraints): I can attach the actuators to the vehicle at run-time using constraints. This works, but the constraints chain can be too long and we can have (in some cases, surely we have) solver convergence problem. So, we need another approach.
2. Vehicle is a MultiBody and actuators are MultiBody: can I atttach a MultiBody actuator base to the MultiBody vehicle base? If I think the overall system as a SINGLE MultiBody, ok, I can do that: only one base and a monolitic representation of vehicle and a single actuator. But what happens if I need modularity? I would like be able to load a kind of MultiBody actuator with its own "simulation" and then attach this one to the vehicle at run-time (so, I can exchange actuators dynamically). Can I "convert" the actuator base in a vehicle link and then rebuild the MultiBody?
3. Vehicle is a RigidBody and actuators are MultiBody: it would be a dream if I could to mix RigidBody and MultiBody, but I think this is not possible because they are two different physic representations. I'm right?
4. Vehicle is a MultiBody and actuators are a chain of RigidBody: useless solution, same problems as 1+3.
Can you enlighten me with some considerations to these points?
Best Regards,
Dario
This is a general question in order to approach the development of an application using MultiBody and/or RigidBody.
Think to a vehicle and some different articulated actuators (excavator, shovel, etc...). I would like be able to model the vehicle physic and then add the actuators to this one, dynamically (at run-time) . I can think to these approaches:
1. Vehicle is a RigidBody and actuators are a chain of RigidBody (using constraints): I can attach the actuators to the vehicle at run-time using constraints. This works, but the constraints chain can be too long and we can have (in some cases, surely we have) solver convergence problem. So, we need another approach.
2. Vehicle is a MultiBody and actuators are MultiBody: can I atttach a MultiBody actuator base to the MultiBody vehicle base? If I think the overall system as a SINGLE MultiBody, ok, I can do that: only one base and a monolitic representation of vehicle and a single actuator. But what happens if I need modularity? I would like be able to load a kind of MultiBody actuator with its own "simulation" and then attach this one to the vehicle at run-time (so, I can exchange actuators dynamically). Can I "convert" the actuator base in a vehicle link and then rebuild the MultiBody?
3. Vehicle is a RigidBody and actuators are MultiBody: it would be a dream if I could to mix RigidBody and MultiBody, but I think this is not possible because they are two different physic representations. I'm right?
4. Vehicle is a MultiBody and actuators are a chain of RigidBody: useless solution, same problems as 1+3.
Can you enlighten me with some considerations to these points?
Best Regards,
Dario