ORGANIZATIONAL LEARNING AGENTS FOR TASK SCHEDULING IN SPACE CREW AND ROBOT OPERATIONS

l'l~is paper explores rescheduling antl reorganizatio~l ahilit,ies of our organizational learning motlel in t.hc following two import,ant. applications in spa.ce: crew task scheduling in a space shuttle/station and task planning for truss construction with multiple spacc rohot,s. Through int,ensive simulat.ions of the above t,wo t,asks, the following experiniental results 1la.vr. Im11 obthiued: (1) Our model provides good feasible sched~llrs quickly in the case of reschedlding, arltl it keeps t,he computational cost for rescheduling low; (2) Plans generated by our nlotlel keep or rc,covt,r efficiency in ta.sks when rohot,s are added, rernovetl, or exchanged among robot. groups; and (3) flrf- integrat,ion of (a) learning mcchanisms, (h) rule Ijastd syst,ems with evolutiona.ry approachrs, antl (c) mult~iagt~l~t approaches is effective in rescl~ecluling/replail~iii~g problems.