“…The other algorithms are to learn action models with incomplete state information [2,[17][18][19][20][21][22][23][24][25][26][27][28][29], which means preconditions and effects can be learned when the state information before and after each action is not fully observed. Yang et al, [17] build an action model learning system-ARMS by using SAT, which can handle the case with incomplete state information, and acquire STRIPS action models.…”