Questionnaire Based Research on Opinions of Visitors for Communication Robots at an Exhibition in Japan

Transcription

1 Quetionnaire Baed Reearch on Opinion of Viitor for Communication Robot at an Exhibition in Japan Tatuya Nomura 1,2, Takugo Taaki 2,3, Takayuki Kanda 2, Maahiro Shiomi 2, Hirohi Ihiguro 2, and Norihiro Hagita 2 1 Department of Media Informatic, Ryukoku Univerity, 1 5, Yokotani, Setaohe cho, Otu, Shiga , Japan 2 ATR Intelligent Robotic and Communication Laboratorie, 2 2, Hikaridai, Seika cho, Soraku gun, Kyoto , Japan 3 Graduate School of Corporate Information, Hannan Univerity, , Amamihigahi, Matubara, Oaka , Japan Abtract. Thi paper report the reult of quetionnaire baed reearch conducted at an exhibition of interactive humanoid robot that wa held at the Oaka Science Mueum, Japan. The aim of thi exhibition wa to invetigate the feaibility of communication robot connected to a ubiquitou enor network, under the aumption that thee robot will be practically ued in daily life in the not o ditant future. More than ninety thouand people viited the exhibition. A quetionnaire wa given to the viitor to explore their opinion of the robot. Statitical analyi wa done on the data of 2,31 repondent. It wa found that the viitor opinion varied according to age; younger viitor did not necearily like the robot more than elderly viitor; poitive evaluation of the robot did not necearily conflict with negative evaluation uch a anxiety; there wa no gender difference; and there wa almot no correlation between opinion and the length of time pent near the robot. 1 Introduction The aim of communication robot i to act in environment with human and ait human through communication with them. Humanoid type robot are conidered to be ueful in thi communication tak by, for example, geturing with their face, arm, and eye in guidance tak for map. One method of implementing communication robot i ubiquitou computing, where robot ue information from enor, not only in the robot themelve but alo in the environment in which they exit [1,2,3]. Thi method aume that all of the object in the environment have their own ID by uing wirele tag ytem [4,5,6]. The mot important characteritic of thi method i it reduced computational cot for identification of environment by robot, which i difficult in cae where each robot mut act alone. Guidance in mueum i conidered an effective application of communication robot uing thee ubiquitou enor network. Although reearch on thi tak M.F. Cotabile and F. Paternò (Ed.): INTERACT 25, LNCS 3585, pp , 25. c IFIP International Federation for Information Proceing 25

2 686 T. Nomura et al. ha been conidered for one robot, it focued on providing information by the robot [7]. From the perpective of communication robot, the interaction between robot and human via enor network information i more important. To invetigate the effectivene of communication robot connected through ubiquitou enor network in guidance tak, an exhibition of humanoid robot, called Robovie [8], wa held at the Oaka Science Mueum 1, Japan, for approximately two month in 24. At thi exhibition, a quetionnaire wa ditributed to viitor to explore their opinion of the robot. Although there ha been ome reearch on pychological evaluation by viitor of robot at cience mueum [9,1,11], thee tudie have been limited to individual impreion of pecific robot behaving alone. The Oaka Science Mueum exhibition focued on interaction between viitor and robot via enor network information in a guidance tak. Thu, the viitor opinion of the robot are aumed to reflect impreion of thi interaction. In particular, the reearch evaluate not only interet in, friendline toward, and perceive effectivene of the robot but alo anxiety toward them. In addition, it focue on relation among thee pychological feature, concrete behavior uch a time pent near the robot, and peronal trait uch a gender and age. Anxiety toward robot and thi feeling relation to behavior and peronal trait are important factor to invetigate when communication robot behave in environment with human and communicate with them [12]. Thi paper give an overview of the communication robot exhibition and then analyze the reult of the quetionnaire gathered there. 2 The Communication Robot Exhibition Thi ection preent an overview of the ubiquitou enor network, the communication robot, the procedure, and the quetionnaire ued at the communication robot exhibition. 2.1 Overview of Sytem The ubiquitou enor network wa contructed on the 4th floor of the Oaka Science Mueum (ee Fig. 1). Thi enor network record viitor behavior, and the information obtained wa ued by the robot to ait viitor in viewing exhibit at the mueum and to encourage their interet in cience and technology. Senor Sytem: In thi exhibition, viitor had wirele tag. Signal from thee tag were detected by uing 2 wirele tag reader. A tag reader can detect ignal from tag within a maximum of 1 m. The interpolation of the trength of ignal detected by everal tag readermake it poible to determine the phyical poition of tag. Eighteen wirele tag reader were hung on the ceiling near exhibit to detect whether viitor tayed near the exhibit. Two tag reader were intalled into the robot. 1

3 Quetionnaire Baed Reearch on Opinion of Viitor 687 Fig. 1. Overview of the Oaka Science Mueum and a Viitor Scene (b) Robovie-M (a) Robovie-II Fig. 2. Two Type of Robot Ued in the Exhibition: (a) Human Sized Robot Robovie II, (b) Small Sized Robot Robovie M In addition, three infrared camera were aigned to detect poition of the robot and four digital camera were aigned to record cene at the exhibition. Each of the camera and tag reader, except for the one aigned to the robot, wa connected to a PC to control information maintained in a databae proceed on a central erver via ethernet. Robot: In the exhibition, two type of communication robot were ued. Figure 2 how the humanoid robot Robovie [8]. Robovie II, hown in Fig. 2(a), i a human ized robot that tand 12 cm tall. It diameter i 4 cm, and it weigh about 4 kg. The robot ha two arm

4 688 T. Nomura et al. (4 2 DOF (degree of freedom)), a head (3 DOF), two eye (2 2 DOFforgaze control), and a mobile platform (two driving wheel and one free wheel). Thi robot ha variou enor, including kin enor covering the entire body, 1 tactile enor located around the mobile platform, an omni directional viion enor, two microphone to liten to human voice, and 24 ultra onic enor for detecting obtacle. It carrie a Pentium III PC on board for proceing enory data and generating geture, including utterance. It i aigned one wirele tag reader. Robovie M, hown in Fig. 2(b), i a mall ized robot that tand 29 cm tall. It ha 22 DOF, which allow it to execute variou geture uch a walking, bowing, and a handtand (ee info/ robot/robovie-m.html). Since the robot doe not have it own function to make utterance, it utterance are performed by the connected PC. 2.2 Procedure Flow of Viitor: Viitor experienced the exhibition a follow. Firt, viitor regiter for their wirele tag at the reception dek at the entrance to the 4th floor (poition A in Fig. 1). At thi tage, their name, age, and birthdate are regitered and input to the ID tag provided them. Then, the regitered name are automatically tranfered into peech information that the robot ue in their utterance to viitor. Viitor are then free to oberve exhibit in the mueum. All of the wirele tag information i recorded in the databae. While viewing the exhibit, viitor interact with robot, each of which ha it own role. One provide guidance of the exhibit while moving alone. Two of them communicate with each other to provide guidance of the exhibit (poition C in Fig. 1). The other robot execute interaction behavior, uch a calling viitor name near the exit (poition D in Fig. 1). When viitor finih viewing the exhibit, they are aked to freely repond to a quetionnaire on their opinion of the robot and the exhibition at the exit (poition E in Fig. 1). The wirele tag are then returned. Role of the Robot: At the exhibition, two Robovie II and two Robovie M were ued. One Robovie II executed exhibit guidance in the mueum while moving about (Fig. 3(a)). It explained the content of exhibit, uch a their hitory. The other Robovie II and one Robovie M executed exhibit guidance while imulating interaction between them by ynchronization via the network (Fig. 3(b)). Specifically, the Robovie M explained an exhibit, the Robovie II aked quetion about it, and finally the Robovie M reponded to the quetion. Furthermore, thee robot interacted with viitor by uing information from the ubiquitou enor network. The remaining Robovie M did not provide guidance but intead interacted with viitor by calling their name baed on viitor tag and regitered information, aying good bye, aking viitor to return their tag, and o on (Fig. 3(c)).

5 Quetionnaire Baed Reearch on Opinion of Viitor 689 (a) (b) Fig. 3. Scene of Interaction between the Robot and Viitor: (a) Guidance by Robovie II, (b) Interaction between Robovie II and Robovie M, (c) Interaction with Robovie M near the Exit (c) Quetionnaire Item: The quetionnaire ued in the exhibition conited of the following tatement. Repondent indicated the degree to which each tatement applie to them by marking whether they (1) trongly agree, (2) agree, (3) are undecided, (4) diagree, or (5) trongly diagree. Item 1 (Interet): I am intereted in the robot. Item 2 (Friendline): I felt friendly toward the robot when I faced them. Item 3 (Effectivene): I find guidance provided by the robot effective. Item 4 (Anxiety toward Interaction): I felt anxiety when the robot talked to me. Item 5 (Anxiety toward Social Influence): I feel anxiety about the poible widepread application of robot to perform tak uch a thoe hown at the exhibition in the near future. The 1t, 2nd, and 3rd item meaure repondent interet in the robot, friendline toward the robot, and evaluation of the robot effectivene, repectively. The 4th and 5th item meaure the repondent anxiety toward interaction with the robot and the ocial influence of the robot, repectively.

6 69 T. Nomura et al. The quetionnaire alo include item on gender and age. The item on age ha even graded anwer (for repondent age categorie from the 1 to 7 ). In addition, the quetionnaire ha an item for freely decribing opinion about the robot and the exhibition. 3 Analyi of Data The communication robot exhibition wa held at the Oaka Science Mueum, Japan, from July to Augut, 24. Thi period included the Japanee ummer holiday. By the end of the two month period, the number of viitor reached 91,17 and the number of viitor who wore wirele tag wa 11,927. The total number of returned quetionnaire wa 3,34, the number of thoe including all of the five item hown in ection 2.2 wa 2,891, and the number of thoe including both the gender item and the age item wa 2,31. Analyi conidering factor of age and gender wa executed for thee 2,31 ample. Finally, the number of quetionnaire that included freely decribed opinion about the robot and the exhibition wa 293. Male Female (N) 1 1% 8 8% 6 4 6% 4% 2% 2 % M F M F M F M F M F Item 1 Item 2 Item 3 Item 4 Item 5 (a) (b) 1% % 6% 4% 2% % Item 1 Item 2 Item 3 Item 4 Item 5 (c) Fig. 4. Ditribution of Repondent and Item Score: (a) Ditribution of Repondent baed on Gender and Age, (b) Ditribution of Item Score baed on Gender, (c) Ditribution of Item Score baed on Age

7 Quetionnaire Baed Reearch on Opinion of Viitor 691 Anwer were cored in revere order of their liting, from 1 ( trongly diagree ) to 5 ( trongly agree). Moreover, the following information wa meaured a a behavior index, baed on tag information from the ubiquitou enor network: T3: Time that viitor tayed within 3 m of the point where Robovie II and Robovie M imulated their communication. The relationhip between thi behavior index and the item core wa alo analyzed. 3.1 Item Score The number of male repondent wa 777 and that of female repondent wa 1,524. Moreover, the number of repondent aged in the 1 wa 349, that in the 2 wa 182, that in the 3 wa 119, that in the 4 wa 519, that in the 5 wa 56, that in the 6 wa 61, and that in the 7 wa 25. Figure 4(a) how the ditribution of repondent baed on gender and age. Thi figure Table 1. Mean Score and Standard Deviation of Item baed on Gender and Age, and Reult of Two Way ANOVA for the Item Score Item 1 Item 2 Item 3 Item 4 Item 5 1 Male Mean (N=144) SD Female Mean (N=25) SD Male Mean (N=64) SD Female Mean (N=118) SD Male Mean (N=287) SD Female Mean (N=822) SD Male Mean (N=215) SD Female Mean (N=34) SD Male Mean (N=19) SD Female Mean (N=37) SD Male Mean (N=31) SD Female Mean (N=3) SD Male Mean (N=17) SD Female Mean (N=8) SD F Value Gender Age Interaction p Value Gender Age Interaction

8 692 T. Nomura et al. indicate that there wa a bia among repondent aged in their 3 and 4, in particular, female in their 3. Figure 4(b) and (c) how the ditribution of the item core baed on gender and age, repectively. Thee figure how that the rate of repondent coring more than 4 on item 1 and 2 were more than 8% and about 7%, repectively. Moreover, the rate of the repondent coring le than 2 on item 4 and 5 were each about 6%. On the other hand, thee figure imply that the ditribution of the item core may differ between age. Table 1 how mean core and tandard deviation of the item baed on gender and age, and the reult of a two way ANOVA for the item core with factor of gender and age. There were tatitically ignificant difference for item 2, 3, and 5 between age. There wa no tatitically ignificant difference between gender. A Tukey pot hoc tet obtained the following reult: The core of item 2 in the 2 group were lower than thoe in the 3, 4, 5 and 6 group. The core of item 3 in the 2 group were lower than thoe in the 1 and 6 group. Moreover, thoe in the 1 group were higher than thoe in the 3 and 4 group. The core of item 5 in the 1 group were lower than thoe in the 2, 6, and 7 group. Moreover, thoe in the 4 group were lower than thoe in the 2 and 3 group. 3.2 Time That Repondent Stayed Near the Robot The behavior index T3 may reflect the repondent interet, friendline, and anxiety toward the robot to ome extent. However, it can alo be influenced by external factor, uch a congetion on the floor. In fact, the number of viitor per day wa widely ditributed during the period (maximum: 3,24, minimum: 767, average: 1,898, median: 1,78), due to the fact that thi period included the Japanee ummer holiday. Thu, the day that more than 2,25 people viited, including ummer holiday, were aumed to be congeted day and the effect of congetion on the behavior index wa analyzed. Firt, a two way ANOVA with factor of the congetion condition and age wa executed. Only the congetion condition had an effect (age: F = 1.186, p =.83; congetion: F =2.46, p =.; interaction: F =.885, p =.55). Next, a two way ANOVA with factor of congetion and gender wa executed. Both congetion and gender had an effect (gender: F =8.111, p =.4; congetion: F =44.93, p =.; interaction: F =1.171, p =.279). Figure 5(a) how the mean value and tandard deviation of T3 on the gender and congetion condition. It wa found that the T3 value of the viitor on congeted day were about 5 ec larger than thoe on non congeted day, and the T3 value of the female repondent were more than 1 ec larger than thoe of the male viitor.

9 Quetionnaire Baed Reearch on Opinion of Viitor C NC C NC A B C D E F G H I J K L Male Female -1 (a) (b) Fig. 5. Mean Value and Standard Deviation of T3: (a) on Gender and Congetion Condition (C: Congetion, NC: Non Congetion, Male C: N = 414, Male NC: N = 363, Female C: N = 699, Female NC: N = 825), (b) on Categorie of Freely Decribed Opinion (A: N = 14, B: N =6,C:N =9,D:N =5,E:N = 26, F: N =5,G:N = 12, H: N =5,I:N =6,J:N =6,K:N = 18, L: N = 1) Table 2. Peaon Correlation Coefficient r between the Item Score and Behavior Index T3 Item 1 Item 2 Item 3 Item 4 Item 5 T3 Item Item Item Item Item Correlation Between Item Score and Behavior Index Table 2 how Peaon correlation coefficient r between the item core and behavior index T3. There were medium level correlation between item 1 3 and between item 4 5. On the other hand, there were few correlation or low level correlation between the group of item 1 3 and that of item 4 5. Moreover, there were few correlation between the item core and behavior index T Freely Decribed Opinion A total of 293 entence expreing opinion on the robot and exhibition were manually claified into everal categorie baed on the content of the entence. Thi claification wa executed by two people who dicued the content of the entence and categorie until they reached a conenu in their claification reult. Finally, 16 categorie were etablihed and each entence wa claified into one of them. Table 3 how thee categorie, the number of entence claified into each category, and example of the entence claified into each category.

10 694 T. Nomura et al. Table 3. Categorie of Freely Decribed Opinion of the Robot and Exhibition, the Number of Opinion Claified into Each Category, and Example of Opinion Claified into Each Category Category N A. Poitive Opinion of the Robot Themelve 23 (Example: I wa glad to have the robot talk me. ) B. Expectation for Robot and Technology in the Future 16 (Example: I would enjoy it if there were more kind of robot. ), C. Poitive Attitude of Children toward the Robot 17 (Example: My child eemed to be glad to be called by the robot. ) D. Deire on Interaction or Touch with the Robot 8 (Example: I wanted to talk with the robot more. ) E. Negative Opinion of Communication with the Robot 59 (Example: The robot utterance were hard to liten to. ) F. Negative Emotion toward the Robot 1 (Example: I felt a little fear toward the robot. ) G. Fear of Children toward the Robot 2 (Example: My child eemed to feel fear toward the robot. ) H. Children Indifference to or Non Interet in the Robot 9 (Example: My child eemed to loe interet in the robot becaue they did not react to the name tag. ) I. Other Diatifaction with the Robot 12 (Example: The robot reaction were lower than what I expected. ) J. Phyical Danger in Interaction with the Robot 7 (Example: The robot arm truck my child. ) K. Poitive Evaluation of the Exhibition 27 (Example: I wa happy becaue I could directly come in contact with the robot. ) L. Critical Requet for the Content of the Exhibition 17 (Example: Pleae prepare more kind of robot. ) O. Other 4 Categorie 68 Categorie A D were poitive opinion of the robot themelve. A correpond to entence expreing poitive opinion and emotion toward the robot appearance, interaction, intelligence, and o on. B correpond to entence expreing expectation and deire for robot and technology in the future. C correpond to entence expreing poitive attitude held by children toward robot, decribed by the children themelve or their parent. D correpond to entence uch a I wanted to interact with the robot more. Categorie E J were negative opinion of the robot themelve. E correpond to entence expreing diatifaction with and negative opinion of the robot function of utterance, recognition, communication, and o on. F correpond to entence expreing negative emotion toward robot, uch a anxiety, fear, and o on. G correpond to entence tating that children felt fear or anxiety toward the robot, a written by the children themelve or their parent.

11 Quetionnaire Baed Reearch on Opinion of Viitor 695 H correpond to entence indicating that children were indifferent to or had no interet in the robot, a written by the children themelve or their parent. I correpond to entence expreing other diatifaction with the robot. J correpond to entence about phyical danger in interaction with the robot, uch a the fact that a robot arm truck the viitor body. Categorie K and L were opinion on the exhibition. K correpond to a poitive evaluation, uch a I would like to viit here again. L correpond to critical requet about the content of the exhibition, uch a type of robot to be exhibited. The other four categorie correpond to entence on diatifaction with external factor not related to the robot or content of the exhibition, uch a congetion of the floor and waiting time for demontration. Thu, thee four categorie were reduced in analyi. Repondent of categorie A D and K were grouped a thoe having poitive opinion, and repondent of categorie E J and L a thoe having negative opinion. The number of poitive opinion and that of negative opinion were 91 (31%) and 134 (45.7%), repectively. The opinion claified into A and K dominated more than half of the poitive opinion. Moreover, category E had the larget number of opinion among the negative opinion and acounted for 44% of the negative opinion. In order to invetigate the relationhip between thee opinion and the time that the repondent tayed near the robot, a one way ANOVA with the opinion categorie wa executed for the behavior index T3. Since the external factor of congetion may influence the analyi, a mentioned in ection 3.2, thi ANOVA wa limited to the repondent on the non congeted day. Figure 5(b) how the mean value and tandard deviation of T3 on the categorie. A a reult, there wa a tatitically ignificant effect of the categorie (F =2.93, p =.2). A Tukey pot hoc tet found that the T3 value of the repondent claified into C were larger than thoe in all the other categorie except for H. 3.5 Dicuion Influence of Age: The reult preented in ection 3.1 how that many viitor poitively evaluated the robot. More pecifically, many viitor had interet in and felt friendline toward the robot. Moreover, many viitor did not feel anxiety about interaction with the robot and their ocial influence. On the other hand, there were difference in thee opinion between age. The reult how that people in their 2 feel le friendline toward robot than thoe in their 3 6, people in their 2 le poitively evaluate guidance by the robot than thoe in their 1 and 6, people in their 1 more poitively evaluate the guidance than thoe in their 3 and 4, people in their 1 feel le anxiety about the ocial influence of the robot than thoe in their 2, 6, and 7, and people in their 4 feel le anxiety about the ocial influence than thoe in their 2 and 3. In other word, younger repondent do not necearily like the robot more than elderly repondent. The above reult imply that the deign of robot hould be changed according to uer age.

12 696 T. Nomura et al. Relationhip to Behavior: The reult preented in ection 3.2 how that ome external factor influence concrete behavior in real ituation, uch a mueum. However, the reult in ection 3.3 how that there i no relationhip between opinion of the robot and the concrete behavior of taying near the robot. Thee reult imply that environmental factor may more trongly affect behavior than pychogical factor in real ituation uch a mueum. Moreover, the reult in ection 3.3 alo how that interet in, friendline toward, and evaluation of effectivene of the robot do not necearily conflict with anxiety toward them. Thee reult imply that robot deign intended to increae effectivene and friendline do not necearily reduce the anxiety felt toward the robot. Attitude of Children Toward Robot: The reult given in ection 3.4 indicate that there are both poitive and negative opinion of the robot and the exhibition on a concrete level. They alo how that there exit everal area of diatifaction with the function of the communication robot, and people, in particular children, may have negative emotion toward the robot at their current level. On the other hand, the reult alo how that there are children who had interet in and friendline toward the robot and indicate that thee children and their parent tay near the robot longer than other. The above reult can be interpreted a follow. In Japan, there are everal type of dicoure on robot, and their effect naturally differ between age group. The reult in ection 3.1 reflect thi. Moreover,manychildren havenever een actual moving robot, although they are affected by everal media. Thi gap may lead to the fear and anxiety toward the robot hown in ection 3.4. If thi interpretation i valid, we can conclude that the deign of robot for children hould be adapted for the image of robot preented in the variou media. Gender Difference: The reult in ection 3.2 reveal a tendency for female to remain near the robot longer than male. However, there i ome doubt a to whether there i a gender difference in behavior toward the robot, a hown in ection 3.2, at leat in the ituation preented in thi reearch. In fact, there wa no gender difference in opinion hown for the item, nor any correlation between them and the behavior index. A a reaon, it can be urmied that many of the viitor were female in their 3 and 4. The period included the ummer holiday, and, a a reult, many female viited the exhibition with their children. In other word, it can be aumed that their children tayed near the robot longer with them and, a a reult, the female appeared to be taying longer. Thi aumption need to be invetigated through another type of data, uch a orbit in which the viitor moved while viewing the exhibit. Such data will be gathered and analyzed in future reearch. 4 Summary Thi paper reported the reult of quetionnaire baed reearch conducted at an exhibition of interactive humanoid robot that wa held at the Oaka Science

13 Quetionnaire Baed Reearch on Opinion of Viitor 697 Mueum, Japan, with the aim of invetigating the ue of communication robot connected with a ubiquitou enor network. More than ninety thouand people viited the exhibition and a quetionnaire wa given to the viitor to explore their opinion of the robot. Statitical analyi wa done for data coniting of 2,31 repondent. It wa found that the viitor opinion of the robot differed according to age, younger people did not necearily like the robot more than elderly people, poitive evaluation of the robot did not necearily conflict with negative evaluation uch a anxiety, there wa no gender difference in opinion of the robot, and there wa almot no correlation between the opinion and the length of time pent near the robot. A future reearch, the relationhip between the viitor opinion of the robot and another behavior index hould be explored. Moreover, there wa a bia of repondent in aembling ample. Although thi bia may be unavoidable in ituation uch a mueum, data from variou type of people need to be gathered for analyi. Acknowledgment Thi reearch wa upported by the Minitry of Internal Affair and Communication. We wih to thank the taff at the Oaka Science Mueum for their highly appreciated cooperation and helpful uggetion: Hideaki Terauchi, Tohihiko Shibata, Koutarou Hayahi, Maaaki Kakio, Taichi Tajika, and Fumitaka Yamaoka. Reference 1. Ng, K.C., Ihiguro, H., Trivedi, M.M., Sogo, T.: An integrate urveillance ytem human tracking and view ynthei uing multiple omni directional viion enor. Image and Viion Computing Journal 22 (24) Ikeda, T., Ihida, T., Ihiguro, H.: Framework of ditributed audition. In: Proc. 13th IEEE Int. Workhop on Robot and Human Interactive Communication (RO MAN). (24) 3. Murakita, T., Ikeda, T., Ihiguro, H.: Human tracking uing floor enor baed on the Markov chain Monte Carlo method. In: Proc. Int. Conf. Pattern Recognition (ICPR). (24) Nihimura, T., Itoh, H., Nakamura, Y., Yamamoto, Y., Nakahima, H.: A compact battery le information terminal for real world interaction. In: PERVASIVE 24. Number 31 in LNCS. Springer (24) Sumi, Y., Matuguchi, T., Ito, S., Fel, S., Mae, K.: Collaborative capturing of interaction by multiple enor. In: Proc. Int. Conf. Ubiquitou Computing (Ubicomp). (23) Schulz, D., Fox, D., Hightower, J.: People tracking with anonymou and id-enor uing rao blackwellied particle filter. In: Proc. Int. Joint Conf. Artificial Intelligence (IJCAI). (23) Burgard, W., Cremer, A.B., Fox, D., Hähnel, D., Lakemeyer, G., Schulz, D., Steiner, W., Thrun, S.: The interactive mueum tour guide robot. In: Proc. Nat. Conf. Artificial Intelligence (AAAI). (1998)

14 698 T. Nomura et al. 8. Ihiguro, H., Ono, T., Imai, M., Maeda, T., Kanda, T., Nakatu, R.: Robovie: an interactive humanoid robot. Int. J. Indutrial Robot 28 (21) Shibata, T., Wada, K., Tanie, K.: Tabulation and analyi of quetionnaire reult of ubjective evaluation of eal robot at Science Mueum in London. In: Proc. Int. Workhop on Robot and Human Interactive Communication (RO MAN). (22) Shibata, T., Wada, K., Tanie, K.: Subjective evaluation of a eal robot at the national mueum of cience and technology in Stockholm. In: Proc. Int. Workhop on Robot and Human Interactive Communication (RO MAN). (23) Shibata, T., Wada, K., Tanie, K.: Subjective evaluation of a eal robot in Burunei. In: Proc. Int. Workhop on Robot and Human Interactive Communication (RO MAN). (24) Nomura, T., Kanda, T., Suzuki, T., Kato, K.: Pychology in human robot communication: An attempt through invetigation of negative attitude and anxiety toward robot. In: Proc. the 13th IEEE International Workhop on Robot and Human Interactive Communication. (24) 35 4