A Decentralized Partially Observable Markov Decision Model with Action Duration for Goal Recognition in Real Time Strategy Games

Algorithm 1

Multiagent joint goal inference based on SIS PF with resampling under Dec-POMDM-T.

Input: particle number , agent team size , resampling threshold .

(1) Set time steps .

(2) For  

(3) sample , and set . % Initialization

(4) End For

(5) For  

(6) For  

(7) If    % Check if joint goal terminate

(8) .

(9) Else

(10) SampleJointGoal.

(11) End If

(12) Observe.

(13) SampleGoalTerminate.

(14) For  

(15) If  

(16) .

(17) TimeDurationUpdate.

(18) Else

(19) SampleActionChange.

(20) ComputeTimeDuration.

(21) End If

(22) SampleActionTermination.

(23) End For

(24) Perform. % Action Perform

(25) End For

(26) For  

(27) Calculate the importance weights

(28) End For

(29) Normalize. % Weight normalization

(30) Calculate , return if ; otherwise resampling

(31) End For