Fuzzy Evaluation Method of Virtual Reality for Urban Landscape Design

Transcription

1 Fuzzy Evaluation Method of Vitual Reality fo Uban Landscape Design Hsiaolin Wang 1, Yufeng Ho 2 and Kuochin Chen 2 1 Depatment of Landscape, Tunghai Univesity, Taichung, Taiwan, whl1435@hotmail.com 2 Gaduate School of Achitectue and Uban Design, Chaoyang Univesity of Technology, Taiwan, hyfach@ms32.hinet.net ABSTRACT: The pupose of this study is to use vitual eality (VR) modeling techniques and fuzzy set theoy to ceate a multi-dimensional and multi-facto fuzzy synthetic evaluation method using VR in Uban Landscape Design. This would allow the geneal evaluation of Uban Landscape Design esults to be enumeated, quantified and computeized. Landscape designes could then eceive immediate feedback on the elative quality of uban landscape design decisions. The Chunghua Road pedestian zone in Taipei City was selected as the subject of this case study. Thee VR landscape design options wee poposed fo evaluation, and the esults showed that the use of VR did stimulate inteest and identification ability among the test subjects. The use of the fuzzy geneal evaluation method to choose between options theefoe helped to incease the confidence of design decisions. Keywods: Uban Landscape Design, Vitual Reality, Fuzzy Synthetic Evaluation 1

2 1. FOREWORD In ecent yeas thee has been a constant flux in ideas when it comes to Landscape Design (Kik 1963, Rapopot 1977, Wang & Ho 1986, Sheppad 1989, Meggs 1995, Bishop 2003). Fom the passive to the active, fom layouts into spaces, fom physical envionments into ecologies and fom compute geneated montages into 3D compute animated models. The apid advances in compute technologies have led to significant changes in the content and diection of Landscape Design. The widespead use of mathematical and compute techniques combined with quantitative analysis has led to moe scientific design decisions and impoved design quality. People s pefeence fo landscaping is detemined by a vaiety of visual sensoy factos such as the plantings, colo, textue, foms, dimensions, spaces and installations. Though the juxtaposition of these factos a unique image is ceated. Due to the vaiety, mutability and complexity of the landscape image conditions, a high level of uncetainty exists. It is this uncetainty that leads to its andomness and fuzziness. As landscape design fo uban aeas is an execise that involves multiple citeia, taditional methods of assessment do not adequately eflect many of the qualitative equiements. Though the use of fuzzy mathematics, it is possible to make the fuzzy concepts, fuzzy identification, fuzzy evaluation and fuzzy decision-making of qualitative desciptions mathematical and quantifiable (Ho and Wang 2000). 2. VIRTUAL REALITY Advancements in Compute Aided Design (CAD) technology has seen landscape design become so detailed as to ival eal-wold imagey. Landscape designes now also favo the use of CAD technology to give uses 2

3 a moe diect undestanding of the oveall landscape design. In ecent yeas, compute technology has pogessed to the stage whee Vitual Reality can be ceated. A thee-dimensional vitual envionment can be simulated using a compute; then physical tools such as digital gloves, 3D glasses and 3D display helmets ae used to povide vey ealistic visual, audio and even tactile simulations. Fo a peson placed in such a vitual envionment it s pactically identical to the eal wold. Evaluation factos that affect vitual eality landscape design include steet functionality, building inteface, space scenaio, steet funitue, planting effect and visual simulation. Due to the inceasing complexity of the envionment, citeia based on just one standad ae now of little pactical value. Decision-making methods fo multiple citeia ae now attacting inceasing attention. Most decision poblems featue multiple people and multiple citeia, with some citeia being had to descibe in pecise numeical values. Landscape design in paticula involves a lot of effect factos that use subjective language-based judgments. To apply the conventional concept of pecision (non-fuzzy), we must have clealy defined ules o boundaies in ode to make judgments. It is, howeve, difficult to econcile ules and naatives using this appoach. Additionally, thee ae many things in the eal-wold that involve a cetain level of fuzziness. To ovecome this bottleneck in conventional ideas, this study used the fuzzy set theoy to popose an evaluation method diffeent fom taditional appoaches. By applying this to the poblem of multi-citeia decision-making in vitual eality landscape design, we have povided a new altenative design eseach method. 3. FUZZY EVALUATION METHOD 3

4 Uban landscape design is a elatively complicated technical system engineeing discipline that ceates a complex system with multiple factos, multiple indicatos and multiple objectives. As the complexity of the system inceases, the uncetainty and impecision in the desciptive system inceases as well. Uncetainties and impessions possess both andomness and tend to be fuzzy as well. People evaluate a subject though easoned judgments. Thei easoning and judgments ae usually fuzzy as well, so the ultimate decision is a fuzzy decision. This study theefoe chose to use the Fuzzy Evaluation Method (Wang, 2003) to evaluate vitual eality poposals in uban landscape design. This povides decision makes with a efeence they can use to detemine the elative meits of poposals. The evaluation method follows the steps listed below: 3.1. ESTABLISH A FACTOR SET (U) FOR THE URBAN LANDSCAPE DESIGN EVALUATION INDEX SYSTEM To evaluate vitual eality poposals fo uban landscape design, one must fist establish a citeia system. This enables an undestanding of the elationships between influencing factos so the system can make the desied evaluation. Taking landscape design elements and visual simulation theoy into account while also efeing to the eseach esults of elevant expets and scholas, the poposed uban landscape design evaluation facto set was as follows: U= {Spatial Function, Building Inteface, Spatial Scenaio, Steet Funitue, Planting Effect, Visual Simulation} 3.2. ESTABLISH A FACTOR WEIGHTING SET (A) The Fuzzy AHP method was chosen fo this study as people s pefeences in uban landscape design vay fom peson to peson. To eflect the impotance of each 4

5 evaluation facto ui, all wee assigned a coesponding weighting ai (i =1,2,,n) with the set of all weightings A =(a 1, a 2,,a n ) efeed to as the facto weighting set. Since factos vay in thei impotance, the weighting of the factos chosen above must be detemined to povide a efeence fo weightings duing evaluation. The weighting set can theefoe be consideed the fuzzy subset of the facto set and can be expessed as: a1 a2 a Α = u u u 1 2 m m 3.3. ESTABLISH THE EVALUATION SET (V) To investigate the evaluatos pefeence fo uban landscape design vitual eality poposals, this study set up a five-level scale fo the evaluatos comments and assessment of the poposals. This esulted in the Evaluation Set V = {Vey poo (v 1 ), Poo (v 2 ), Aveage (v 3 ), Good (v 4 ), Vey Good (v 5 )} 3.4. ESTABLISH THE FUZZY RELATIONSHIP MATRIX (R) The subject is evaluated accoding to the i-th facto (ui) in the facto set. Its degee of membeship to the j-th element (vj) in the evaluation set is ij. The esult of the evaluation fo the i-th facto ui can be expessed as the following fuzzy set: i1 vi2 i Ri = v v v 1 2 n n Ri is efeed to as the single facto evaluation set, Ri = (i 1, i 2,,i n ) Combine the degee of membeship behavio of each single facto evaluation set to fom a fuzzy elationship matix: R R R R 1 2 m = m m2 1n 2n mn 5

6 The single facto evaluation set can actually be teated as the evaluation given fo each facto in the poposals deived fom facto set u and the evaluatos undestanding of the poposals desciptions. The evaluation method is based mainly on the fuzzy ate scoe with language meaning defining the scoe intevals. Each inteval epesents one tiangula fuzzy numbe with the midpoint of the inteval being the expet s chosen answe and the lowe/uppe values being ange of ate scoes acceptable to the expet. As the integation of similaity degee at this stage is moe staightfowad than the integation of facto weightings, the ate scoes in the scale ae mathematically significant. The method used to deive the weightings was theefoe not necessay fo the factos FUZZY EVALUATION OF PROPOSALS (B) Jointly evaluate the influence of all factos to aive at the coect evaluation esults. Taking the weights of each facto into account in the single facto evaluation matix will allow the combined influence of all factos to be epesented in a easonable manne. The fuzzy evaluation can theefoe be expessed as: B = A R = = ( b, b, b ) 1 2 ( a, a, a ) n 1 2 m R R R 1 2 m = m m2 1n 2n mn Hee B is efeed to as the fuzzy evaluation set. This defines the concete evaluation esults fo the uban landscape design vitual eality poposals. 4. CASE STUDY The Zhonghua Steet Ximen Pedestian Aea in Taipei City s Wanhua Distict (Fig. 1) 6

7 was selected as the case study fo empiical eseach. Pedestianization can be intepeted as a kind of spatial pactice on the epesentation of space. The pedestianization effots made to Ximending aound 1990 left much oom fo impovement. The No. 6 Ximen exit s suoundings fo example fomed both the enty squae and a landmak fo waiting so was not use fiendly. The defects included inappopiate tansition buffe space, lack of ovehead cove against sun o ain and lack of appopiate waiting space. The cycle path netwok though the Zhonghua Steet pedestian zone was clealy maked out and thee ae no special steet funitue designs installed. The oveall esult was one of unbelievably atocious building inteface and spatial quality. Compehensive impovement pojects wee theefoe poposed. Ximen Pedestain Aea Special Scenic Spot Main Road The Study Aea Figue 1:-Location of empiical study 7

8 4.1. PROPOSALS Fist, the thee pedestian zone landscape design candidate poposals wee modeled in 3D using AutoCAD and 3D Max 7.0. The models wee then impoted using Vitual Reality tools into the Vitools Dev 3.5 softwae to ceate thee poposals VR space (Table 1). The landscape designes could then immese themselves in the ealistic vitual envionment to inteact with and expeience thei design. Table 1: Vitual Reality Candidate Poposals Ximen Landscape Design Impovement Poposals Pesent Scene Poposal A Poposal B Poposal C Design Concept The defects included inappopiate tansition buffe space, lack of ovehead cove against sun o ain and lack of appopiate waiting space. The cycle path netwok though the Zhonghua Steet pedestian zone was clealy maked out and thee ae no special steet funitue designs installed. The oveall esult was one of unbelievably atocious building inteface and spatial quality. Compehensive impovement pojects wee theefoe poposed. Geen vegetation is planted to beautify the landscape. This ceates a eceation space with landscaping and natual elements that pedestians will find moe welcoming. To maintain pedestian visibility, vegetation should use tees with highe cowns athe than shot close-gown bushes. This avoids the plantings impacting advesely on sceney and public safety. The design establishes a complete and contiguous pedestian space. Pedestian taffic is encouaged fo destinations within walking distance though a tightly defined envionmental famewok and efficient spatial layout. Tees, flowes and plants help to define the space s diection enhancing the visual quality of sevice. Ceate a quality pedestian envionment with space fo the public to play and elax. Vaiety of visual landscape is enhanced though the use of vaiable paving colos and mateials. Emphasis is given to open public spaces to impove pedestian comfot. Steet funitue is neatly aanged and elegant lighting povided. The design also calls fo moto vehicle-pedestian sepaation system and cycling space to be installed. 8

9 4.2. ESTABLISHING THE FACTOR SET OF THE EVALUATION INDEX SYSTEM This study consulted liteatue elating to uban landscape evaluation and visual simulation factos (Litton, 1974; Jones & Jones, 1977; Kaplan & Kaplan, 1978; Ho & Wang, 1986, 2000; Sheppad, 1989; Booth, 1990, Faenc et al, 2000; Bishop & Rohmann, 2003; Fukahoi & Kubota, 2003; Boian et al, 2004) to daw up a peliminay evaluation index system (Table 2). Laye 1 contained 2 items: Envionmental Context and Landscape Visual; Laye 2 contained 6 items including Spatial Composition; Laye 3 contained 28 items including System. Table 2: Index System fo Evaluation of Uban Pedestian Zone Landscape Design System Steet Functionality Functionality Safety Envionmental Context Building Inteface Convenience Decoativeness Uniqueness Intuitiveness Unifomity Spatial Scenaio Guidance Comfot Leisue Inteest Index System fo Evaluation of Uban Pedestian Zone Landscape Design Envionmental Visual Steet Funitue Planting Effect Visual Simulation Richness Completeness Aesthetics Divesity At Tansmissivity Continuity Vaiety Hamony Ode enclosue Tuthfulness Repesentation Liveliness Claity Identification 4.3. FILTERING OF EVALUATION FACTORS AND EXPERT WEIGHTING The setting of each facto s weighting used FAHP softwae to calculate the 9

10 Table 3: Index System fo Evaluation of Uban Pedestian Zone Landscape Design Taget Laye Level 1 Laye 2 Evaluation Facto Index System fo Evaluation of Uban Pedestian Zone Landscape Design (1) Envionmen tal Context (0.4) Envionmen tal Visual(0.6)* Steet Functionality (0.289) Building Inteface (0.485) * Spatial Scenaio (0.226) Steet Funitue (0.297) Planting Effect (0.203) Visual Simulation (0.5) * ( ) Paentheses indicate expet weighting * indicates highest weighting fo that facto set Relative weighting System (0.319) Safety (0.280) Convenience (0.401) * Decoativeness (0.432) * Uniqueness (0.220) Intuitiveness (0.090) Unifomity (0.258) Comfot (0.430) * Leisue (0.333) Richness (0.238) Completeness (0.288) Aesthetics (0.330) At (0.382) * Continuity (0.293) Vaiety (0.108) Hamony (0.252) Ode (0.347) * Tuthfulness (0.108) Repesentation (0.305) * Liveliness (0.235) Claity (0.244) Identification (0.109) Consistency Index (C.I.) and Consistency Ratio (C.R.); when C.I. < 0.1, the degee of consistency was viewed as satisfactoy; when C.R. < 0.1, the ate scoe within the matix was usable. The deived evaluation factos and thei weightings ae shown in Table VIRTUAL REALITY PROPOSAL SURVEY RATING In this study, the fuzzy ating scoes on the poposals fom expets and non-expets wee integated though similaity to deive the fuzzy ating scoe intevals fo the factos in each poposal as shown in Table FUZZY EVALUATION OF VIRTUAL REALITY PROPOSALS 10

11 Table 4: Table of Fuzzy Rating Scoes fo Each Facto in Each Poposal Poposal A Poposal B Poposal C System (11.92,19.97,28.02) (8.89,16.93,24.98) (13.33,21.38,29.42) Safety (8.61,16.95,25.28) (15.27,22.90,30.53) (8.08,16.27,24.46) Convenience (13.37,21.13,28.90) (11.38,19.57,27.76) (18.32,25.25,32.19) Decoativeness (13.86,21.77,29.68) (4.73,12.64,20.55) (12.18,20.52,28.85) Uniqueness (8.61,16.94,25.28) (9.73,17.92,26.11) (5.85,14.18,22.51) Intuitiveness (13.85,21.90,29.94) (11.32,19.65,27.99) (12.79,20.69,28.60) Unifomity (18.35,25.55,32.76) (10.77,19.10,27.44) (13.32,21.23,29.14) Comfot (13.33,21.52,29.71) (11.39,19.44,27.49) (13.03,20.94,28.85) Leisue (14.72,22.63,30.54) (9.17,17.50,25.84) (8.33,16.67,25.00) Richness (14.12,22.17,30.22) (9.18,17.51,25.84) (9.44,17.49,25.54) Completeness (14.11,21.88,29.65) (7.79,16.12,24.46) (7.23,15.56,23.89) Aesthetics (10.00,18.33,26.66) (7.22,15.41,23.60) (6.12,14.45,22.78) At (12.23,20.28,28.33) (8.06,16.40,24.73) (7.78,15.96,24.15) Continuity (11.40,19.73,28.06) (10.56,18.61,26.65) (12.21,19.98,27.75) Vaiety (8.33,16.67,25.00) (10.22,17.99,25.75) (6.64,14.42,22.19) Hamony (10.59,18.64,26.69) (11.92,19.97,28.02) (13.01,21.06,29.11) Ode (9.73,17.92,26.11) (10.56,18.47,26.38) (11.07,19.12,27.16) Tuthfulness (11.92,20.25,28.59) (11.36,19.27,27.18) (11.08,19.27,27.46) Repesentation (13.00,21.19,29.38) (6.94,14.85,22.75) (10.56,18.46,26.37) Liveliness (12.98,21.03,29.08) (13.27,21.46,29.65) (12.18,19.67,27.16) Claity (13.55,21.75,29.94) (4.72,12.91,21.10) (16.93,23.88,30.82) Identification (13.60,21.64,29.69) (11.96,19.87,27.78) (3.34,18.34,26.87) Note: In (a, b, c), a epesents the left limit of the tiangula fuzzy numbe, b epesents the midpoint value of the tiangula fuzzy numbe and c epesents the ight limit of the tiangula fuzzy numbe. Afte deiving the atings fo each facto in each poposal, the oveall evaluation of the poposals can be caied out to identify the poposal with the highe ating. To make the calculation pocess and esults cleae, it was decided that the calculations would use the middle values of the ating intevals. The calculation method fo the combined evaluation sets is as follow: B=A R =(0.046,0.040,0.058,0.105,0.053,0.022,0.063,0.049,0.038,0.027,0.043,0.049,0.057,0.030,0.011,0.026, 0.035, 0.027, 0.076, 0.059, 0.061, 0.027) =(20.754, ,19.183) 19.97, 16.93, , 22.90, , 19.57, , 12.64, , 17.92, , 19.65, , 19.10, , 19.44, , 17.50, , 17.51, , 16.12, , 15.41, , 16.40, , 18.61, , 17.99, , 19.97, , 18.47, , 19.27, , 14.85, , 21.46, , 12.91, , 19.87,

12 The esults of the calculation showed that Poposal A ( points) was slightly supeio to Poposal C ( points) and also Poposal B ( points). The fuzzy evaluation esults of the factos in each poposal showed that out of the 22 factos evaluated, Poposal A was supeio to Poposal C on 16 factos. The biggest diffeence among these was Completeness with a diffeence of appoximately 6.32%. This showed that fo landscape visual, Poposal A s use of steet funitue to define the steet s chaacte was supeio to Poposal C in tems of completeness. Poposal A was supeio to Poposal B on 16 factos as well, with the biggest diffeence being Decoativeness with a diffeence of 9.13%. This showed that fo the envionmental context, Poposal A s use of decoativeness to enhance/beautiful the exteio of the building inteface was supeio to Poposal B. Poposal C was supeio to Poposal B in 12 factos, with the biggest diffeence being Claity whee the diffeence was 10,97%. This showed that in landscape visual, Poposal C s content, details and oveall pesentation in its vitual eality space was moe easily discened than Poposal B. Poposal C was bette than Poposal A in 6 factos, the biggest diffeence being in Hamony whee the diffeence was about 2.42%. This showed that fo landscape visual, Poposal C s planting effect achieved a bette hamony in the steetscape than Poposal A. Poposal B was supeio to Poposal A in 6 factos, the biggest diffeence being in Safety whee the diffeence was 5.95%. This showed that in envionmental context, Poposal B s steet functionality offeed bette safety within the pedestian space than Poposal A. All thee poposals had simila scoes fo Liveliness. This might be due to the use of vitual eality tools fo visual simulation of uban pedestian zone landscape design. VR enhanced the design elements shape, colo, appeaance and textue, stengthening the impession left on the audience. The simila scoes may theefoe be due to the stong impession left and the boad acceptance of the VR-based pesentation fomat. 12

13 5. CONCLUSIONS AND RECOMMENDATIONS 5.1. CONCLUSIONS 1.Landscape design fo uban pedestian zones is a fuzzy decision making question that involves making a judgment fom numeous fuzzy factos. A new multiple objective decision-making method that uses fuzzy math theoy and methods was shown to be feasible. Using this evaluation method, the eliability of decision-making fo uban pedestian zones in landscape design was impoved. 2.This method can be combined with compute-aided design technologies to establish a combined system analysis model that has both qualitative and quantitative chaacteistics. This pomotes the automation of landscape design fo uban pedestian zones. 3.Using the system analysis method to investigate the elationships between influencing factos, layes and component factos in landscape design fo uban pedestian zones so as to establish an evaluation model is feasible. 4.Landscape design fo uban pedestian zones is an extemely detailed undetaking. To accomplish the poject objectives, thoough planning by using VR to ceate a epesentation of the design can help impove efficiency and esults RECOMMENDATIONS 1.Continue to develop and efine this model so it can be tanslated into softwae fo actual use in landscape design fo uban pedestian zones. Thee exists a eal equiement fo quantification standads to be defined fo factos so evaluation can be made simple and moe objective. 2.Ceate modules and databases using VR technology. Also set up an expet 13

14 system to help with designes analysis and assessments so landscape design fo uban pedestian envionments can be made moe scientific and pecise. It will also allow the designe to povide the geneal public with a coss-platfom method of communication. 3.Incopoate landscape elements natual ambient factos such as light, wate and sound into the study. This will add weight to the index system, concept and decision-making in landscape design fo uban pedestian zones, making them moe compehensive. REFERENCES: Booth, N. K., (1990) Basic Elements of Landscape Achitectual Design, Waveland Pess, USA. Bishop, I.D., & Rohmann, B., (2003) Subjective esponses to simulated and eal envionments: a compaison, Landscape and Uban Planning, Austalia, Volume 65, pp Boian, R. F., & Budea, G.C., (2004) Deutsch, J.E. and Winte, S. H.,Steet Cossing Using a Vitual Envionment Mobility Simulato, pp.1-7. Faenc, N, & Musse, S. R., (2000) Schweiss,E.,A Paadigm fo Contolling Vitual Humans in Uban Envionment Simulations, In Applied Atifical Intelligence Jounal, Volume 14, pp Fukahoi, K, & Kubota, Y, (2003) The ole of design elements on the cost-effectiveness of steetscape. Ho, Y.F. & Wang H.L., (1986) Landscape Planning of the Fist Nuclea Powe Plant Aea in Taiwan, TPC. Ho Y.F., & Wang H.L. & Yao S.C., (2000) A Study on the Fuzzy Evaluation Method of Colou Scheme fo Uban Building, Jounal of Design, Volume 5, Numbe 2, pp.9-33, Taiwan Jones and Jones, (1977) Visual Resouce Management fo Highway, Washington D.C.:USDT, Fedeal Highway Administation, National Highway Institute and Office of Envionmental Policy. Kaplan, S. & Kaplan, R., (1978) Humanscape: Envionments fo People, Ann Abo, MI: 14

15 Ulich s. Litton, R.B, (1974) Visual Vulneability of Foest Landscape, Jounal of Foesty, July 84, pp Sheppad, S.R.J., (1989) Visual Simulation:A Use s Guide fo Achitects, Enginees and Plannes, New Yok:Van Nostand Reinhol. Wang, H.F., (2003) Multiciteia Decision Analysis, Fom Cetainty to Uncetainty, pp