Page - 183 - in Intelligent Environments 2019 - Workshop Proceedings of the 15th International Conference on Intelligent Environments

correspond to the user’s budget, available time for the tour and the initial stamina, re- spectively.Weassume that theuserwill set thesevaluesmanually. rn canberepresented byEq. (1). rn=r0− n−1 ∑ i=0 [CR(xi)+moveCR(xi,xi+1)]−CR(xn) (1) where rn=(mn,tn,sn)T CR(x)=(CMx,CTx,CSx)T moveCR(x,x′)=(moveCMx,x′,moveCTx,x′,moveCSx,x′)T Here,CR(x)denotes the resourceconsumptionat spotxandmoveCR(x,x′)denotes the resource consumption whilemoving from spot x to x′.CMx,CTx andCSx denote the consumptionofmoney, time and stamina for enjoying sightseeing at spot x, respec- tively.moveCMx,x′,moveCTx,x′ andmoveCSx,x′ denote the consumption ofmoney, time andstaminaformovingfromspotx tox′, respectively.Weassume that theseareconstant valuesgiven inadvance. We assume that satisfaction denoted by c is determined by the tour route, the stay timeat eachspot and theenvironmental conditionat the spot and route.Then,we repre- sentcbyEq. (2). c(X)= n−1 ∑ i=0 [SAT(xi)+moveSAT(moveCT(xi),xi,xi+1)]+SAT(xn) (2) Here,SAT(x)denotessatisfactionobtainedatspotx,andmoveSAT(moveCT(x),x,x′) denotes satisfactionobtainedwhilemoving from spotx tox′. Weassume that thesatisfactionobtained isalwaysapositivevalue.Theobjectiveof the problem is tomaximize both remaining resources and satisfaction.Then, the objec- tive function is representedbyEq. (3). maximizemn(X),tn(X),sn(X),c(X) (3) 4. GA-basedAlgorithm toDeriveDiverseTourRoutes Theproblem inSect. 3 is anNPhardproblem since it implies themulti-objectiveknap- sackproblem (knownasanNP-hardproblem) as a special case, soweproposeaheuris- tic algorithm to solve it inpractical time. In this section,first,wedescribe thecodingof solutions operated in our algorithm, then describe genetic operators includingmutation andcrossoverused in theproposedalgorithm. Y.Hiranoetal. /AMethod forGeneratingMultipleTourRoutes 183