March 20, 2020
Updating A*
Different Pathfinding Logic for Different Agent Types
A* Current System
The A* Pathfinding currently performs the same calculations for every agent in order to determine its path. Every unit then takes the same path assuming they have the same starting and target positions. We are looking to adjust this so there are various types of agents which with different affinities to certain objects in the environment. This then needs to be incorporated in the path determination so different types of agents will takes different paths, even when using the same starting position and same target position.
Differentiating Pathfinding for Different Types of Agents
We want to add more parameters to the grid nodes besides flat cost that influence the agents’ paths. These parameters also need to be able to influence different agents’ paths differently (i.e. the parameters may give a node a cost of 10 for one agent, but 20 for another).
Approach
The initial needs for this system are the addition of parameters to the nodes of the A* grid and parameters that correlate with these on the agents themselves. With this in place, we are just going to add a reference pass of the individual agents themselves to the eventual path calculation so it can also factor in these additional values.
The actual steps for this concept are laid out here:
– Individual agents need to have data within them that determines how they act
– Pass the UnitSimple instance itself through PathRequestManagerSimple.RequestPath along with the current information (its position, its target position, and the OnPathFound method)
– In PathRequestManagerSimple script, add UnitSimple variable to the PathRequest struct
– Now, RequestPath with added UnitSimple parameter can also pass its UnitSimple reference to its newly created PathRequest object
– Then, TryProcessNext method is run
– Here, the UnitSimple reference can be passed on to PathfindingHeapSimple through the pathfinding.StartFindPath() method call
– Add UnitSimple parameter to StartFindPath method in PathfindingHeapSimple class along with existing parameter (start position of path request, target position of path request)
– PathfindingHeapSimple can now use variables and factors of UnitSimple when determining and calculating the path for that specific agent
– Finally, returns its determined path as a set of Vector3 waypoints to the agent (effectively stays the same process)
Next Steps
This is just a conceptual plan, so I still need to implement it to see if it works and how it runs. I will also be looking to add some parameters to the nodes and agents, so I can start differentiating various types of agents to test the system.