/// <summary>
/// Does an optimised IDA* search
/// </summary>
/// <param name="startModel"></param>
/// <param name="goal"></param>
/// <param name="maxDepth"></param>
/// <returns></returns>
public static List <Action> planActions(WorldModel startModel, Goal goal, int maxDepth, int ttSize)
{
List <Action> path = new List <Action>();
cutoff = startModel.estimateHeuristic(goal); // set heuristic of startModel as optimisation cutoff
TranspositionTable transpositionTable = new TranspositionTable(ttSize);
while (cutoff >= 0 && cutoff != int.MaxValue) //TODO: passiert >=0 überhaupt?
{
//get actionplan
path = doDepthFirst(startModel, goal, transpositionTable, maxDepth);
if (path != null)
{
return(path); //if a valid plan was found, return it
}
}
return(null);
}
/// <summary>
/// IDA* iteration
/// </summary>
/// <param name="worldModel"></param>
/// <param name="goal"></param>
/// <param name="transpositionTable"></param>
/// <param name="maxDepth"></param>
/// <returns></returns>
static List <Action> doDepthFirst(WorldModel worldModel, Goal goal, TranspositionTable transpositionTable, int maxDepth)
{
//storage for models at each depth ans corresponding actions and their cost
WorldModel[] models = new WorldModel[maxDepth + 1];
Action[] actions = new Action[maxDepth];
float[] costs = new float[maxDepth];
//initialize
models[0] = worldModel.copy();
int curDepth = 0;
//keep track of minimal cutoff for optimization
int smallestCutoff = int.MaxValue;
//iterate until there's no more possibilities at depth 0
while (curDepth >= 0)
{
if (curDepth == 0)
{
smallestCutoff = int.MaxValue;
}
//check wether goal is reached
if (goal.isFulfilled(models[curDepth]))
{
List <Action> ret = new List <Action>();
for (int i = 0; i < actions.Length; ++i)
{
if (actions[i] == null)
{
break;
}
ret.Add(actions[i]);
}
return(ret); //return current plan
}
//if maximum depth is reached
if (curDepth >= maxDepth)
{ //continue in lower depth
--curDepth;
continue;
}
//calculate total cost of current plan
float cost = models[curDepth].estimateHeuristic(goal) + costs[curDepth];
//check if we need to prune based on cost
if (cost > cutoff)
{
//
if (cutoff < smallestCutoff)
{
smallestCutoff = cutoff;
}
//drop back
--curDepth;
continue;
}
//otherwise try the next action
Action nextAction = models[curDepth].nextAction();
if (nextAction != null)
{
//copy current model
models[curDepth + 1] = models[curDepth].copy();
//apply action
actions[curDepth] = nextAction;
models[curDepth + 1].applyAction(nextAction);
costs[curDepth + 1] = costs[curDepth] + nextAction.cost;
if (!transpositionTable.has(models[curDepth + 1]))
{
++curDepth;
transpositionTable.add(models[curDepth], curDepth - 1);
}
else
{
transpositionTable.add(models[curDepth + 1], curDepth);
}
}
else
{
--curDepth;
}
}
cutoff = smallestCutoff;
return(null);
}