Page 117 - The ROV Manual - A User Guide for Remotely Operated Vehicles 2nd edition

Collision shapes
Start
FIGURE 4.9
Object AABBs
Collision data
Overlapping pairs
Physics pipeline
Contact points
Time
World transforms velocities
Dynamics data
Constraints Mass (contacts,
4.2 Simulation 105
inertia joints)
End
Apply Predict gravity transforms
Forward dynamics computation
Compute Detect AABBs pairs
Compute contact points
Collision detection computation
Setup Solve Integrate constraints constraints position
Forward dynamics computation
Physics engine logic progression for predicting, modeling, and rendering simulated objects.
Proper joint configuration and constraint solver parameters are essential as well in order to sim- ulate believable manipulators and cables. Complex scenario logic (including triggers) is imple- mented in a similar fashion. Collision models that have been tagged as sensors can report when they have collided with their corresponding target sensors. But any resulting actions must be speci- fied by the programmer or scenario developer (Figure 4.9). Also, in certain situations the developer may wish to override the physics engine and manually position objects or joints (such as in a posi- tionally controlled mechanism).
When considered from this perspective, the physics engine is simply one of many tools that aid developers and programmers in creating realistic and plausible simulations. It is not a “magic box” that can instantly provide a robust ROV simulator without configuration or effort. Much work has to be done to determine and generate the correct forces for each object. Scenarios often must be optimized after initial design to remove excess contact points and redundant calculations. Decisions about certain trade-offs must be made such as the choice between using an iterative constraint solver (that is faster) or a direct solver (that is more accurate). Utilized properly, the physics engine can enable a faster scenario development process, thus reducing the amount of divergence from real-world behavior. This increases the resiliency of the simulation while permitting the level of flexibility developers require in order to implement convincing scenarios.
4.2.3 The future
The future of ROV simulation is here. Engineering analysis tools are bringing together various rele- vant technologies to enhance realism. Developers see the interactivity of critical elements as inputs to real-time human-in-the-loop systems such as ROVs. These allow for a method to visualize tech- nologies in context (Figure 4.10).