The second development phase for Pathfinder 2011 is underway. We’ve already made good progress on waypoint scripting and we’ve also made some changes to enable parallel processing.
The waypoint scripting feature will make it possible to send occupants to arbitrary points within the model. It will also be possible to make occupants wait near a location before moving on to the next waypoint. By combining the actions, users can include additional behaviors (e.g. movement to areas of refuge, firefighting actions) that aren’t possible with the current system – which only permits exiting and an optional initial delay.
The following video shows an early preview of the waypoint scripting feature. The agent that starts in the right side of the model moves against the evacuating crowd to go to a waypoint on the left side of the navigation surface. After a short scripted wait time, the agent then exits the model to end the simulation.
The most computationally expensive element of a Pathfinder simulation is the movement calculation step. We’ve made improvements to this code that enable localized parallel processing. In many of our test problems, quad core workstations are now able to run simulations in about half the time.