IPSJ SIG Technical Report Vol.2013-SLP-97 No /7/27 1 2,1 1 ( ) ( ) Phyno 1 (Phyno) PC Evaluation of Superiority of Robot Agent in Spoken Dialog

Size: px

Start display at page:

Download "IPSJ SIG Technical Report Vol.2013-SLP-97 No /7/27 1 2,1 1 ( ) ( ) Phyno 1 (Phyno) PC Evaluation of Superiority of Robot Agent in Spoken Dialog"

なおつなかわ
5 years ago
Views:

1 1 2,1 1 ( ) () Phyno 1 (Phyno) PC Evaluation of Superiority of Robot Agent in Spoken Dialog Interface Akira Tamagawa 1 Kazumasa Yamamoto 2,1 Seiichi Nakagawa 1 Abstract: The target of speech input can be an agent on the screen (virtual person or creature) or robot (real). When considered from the standpoint of easiness of talk and friendliness, a real robot would be better on the physicality. In this study, we built a dialog system that was set to talk with the pet robot Phyno. The task of this system is to guide rooms in the building, where the location is assumed at the front of the elevator on the first floor of our university building C. In order to compare the impression of the user due to the target s devices in the utterance, we carried out the comparison of the impression for the system with the robot, computer display, and mobile display. From the experimental results, the benefit of physicality of the robot was demonstrated. We also designed the system to display the route on mobile devices and confirmed in the experiment the effect of the robot facing the user. In these experiments, it was possible to obtain a high evaluation rate in such friendliness. Keywords: spoken dialog system, robot, agent, target device, task oriented dialog Toyohashi University of Technology 2 Toyota National College of Technology ( ) POMDP (Partially Observable Markov Decision Process)[1] [2] ( ) ( ) [3] 1

2 [4], [5], [6], [7] [4] [5] [7] [8] CG Autonomous Systems Lab Robox [9] Robox Swiss National Exhibition expo.02 LED 1 Yes/No ( ) Phyno[10] C 1 C [11]

1 [%] 98.7 94.9 2 [%]() 2 Phyno 50cm 100cm / ±15 ±15 ±15 ±15 ±15 ±30 2ch 93.7 - - - - - 3ch 47.2 96.8 64.5 75.6 70.1 96.6 96.7 93.4 92.3 97.0 94.7 99.

3 1 [%] [%]() 2 Phyno 50cm 100cm / ±15 ±15 ±15 ±15 ±15 ±30 2ch ch Phyno[10] 2 (SONY ECM-88B) Phyno Samsung Galaxy Nexus Android OS LAN 2.3 (VAD: Voice Activity Detection) (TDOA: Time Difference Of Arrival) Phyno Phyno TDOA 2 (L,R) TDOA 2 C 1 3 [%]() 50cm 100cm / ±15 ±15 ±15 ±15 ±15 ±30 2ch ch IPA100 ( ) 9 3 (A B C) 2 30 ( ) 2 (50cm 100cm) TDOA (L,R) TDOA 2,3 3ch 2ch % 30 99% 15 82% 30 94% cm 100cm L,R p LR L,C p LC R,C p RC GCC-PHAT L,C 3

4 4 (50cm 100cm) [%] GCC-PHAT L,R () L,R L,C R,C L R p mn 1 SVM(Support vector machine) 60% SVM SVM SVM SV M light version 6.02[12] RBF(Radial Basis Function) ( 10) % 79% % 68% VAD MFCC(Mel-Frequency Cepstral Coefficient) (CFG: Context Free Grammar) SPOJUS[13][14] ( ) 2.5 Phyno GalateaTalk[15] jagtalk[16] 3. ( C 1 ) 3.1 (Phyno) PC 3 PC ( ) 3 PC LAN C 4

5 3 5 ( ::PC : ) (5 ) 6 ( ) 5 6 5

6 7 a) b) (1) x : x={( A PC ) + ( A ) + ( A )}/3 (2) x : A PC x A,P C =( A PC ) x (3) :PC = 1 N xi,p C 1 PC CG WebSocket[17] LAN 6

7 10 7 ( ) : 2:3: 4:5: ( ) ( ) 8 1: 2: 3: 4:5: 7

a) b) 11 8 (5 ) 3.5 4.5 4.0 3.5 3.8 3.5 3.8 3.7 4. Phyno (Phyno) PC [1] Jason D. Williams and Steve Young. Partially observable Markov decision processes for spoken dialog systems. Comput.

8 a) b) 11 8 (5 ) Phyno (Phyno) PC [1] Jason D. Williams and Steve Young. Partially observable Markov decision processes for spoken dialog systems. Comput. Speech Lang., Vol. 21, No. 2, pp , [2],,... SLP,, Vol. 2012, pp. 1 8, [3]. IoT., Vol. 54, No. 7, pp , [4],.., [5],.. HAI 2011, [6],,,,,.. D Vol.J95-D No.10 pp , [7],, , pp , [8],,,,,,.., SP , [9] Roland Siegwart, Kai O. Arras, Samir Bouabdallah, Daniel Burnier, Gilles Froidevaux, Xavier Greppin, Björn Jensen, Antoine Lorotte, Laetitia Mayor, Mathieu Meisser, Roland Philippsen, Ralph Piguet, Guy Ramel, Gregoire Terrien, and Nicola Tomatis. Robox at Expo.02: A Large-Scale Installation of Personal Robots. Robotics and Autonomous Systems, Vol. 42, pp , [10] Phyno. [11] [12] SVM-Light Support Vector Machine. svmlight.joachims.org/. [13],. SPOJUS-SYNO. SP93-20, [14] SPOJUS-SYNO. SPOJUS/. [15] GalateaTalk. projects/galateatalk/. [16] jagtalk [ja.nishimotz.com]. project:jagtalk. [17] WebSocket. 8

& Vol.5 No (Oct. 2015) TV 1,2,a) , Augmented TV TV AR Augmented Reality 3DCG TV Estimation of TV Screen Position and Ro

& Vol.5 No (Oct. 2015) TV 1,2,a) , Augmented TV TV AR Augmented Reality 3DCG TV Estimation of TV Screen Position and Ro TV 1,2,a) 1 2 2015 1 26, 2015 5 21 Augmented TV TV AR Augmented Reality 3DCG TV Estimation of TV Screen Position and Rotation Using Mobile Device Hiroyuki Kawakita 1,2,a) Toshio Nakagawa 1 Makoto Sato