打楽器とロボットとの合奏のための 結合振動子モデルに基づく打撃時刻 ...

RJ200H2. Weinberg , , 4 [1]. , . ,. , . Otsuka , [5]. , , .3. , , . ., . , .3·1 , . , φ(t) , . .φ(t) = (φ 0 + 2πt/T osc ) , (1), t , T osc , φ 0 . , 2π . (1) , .dφ 1= ω 1 , (2)dt, ω 1 . (2) , ω 1., 2 . , (2) 2π . , dφ 1= ω 1 + K 1 Q(φ 2 − φ 1 )dt(3)dφ 2= ω 2 + K 2 Q(φ 1 − φ 2 ),dt(4), φ 1 , φ 2 , Q 2π , K 1 , K 2 , ω 1 , ω 2 , . Q sin [10] , . Attractor(a) K 1 = K 2 = 1Fig.2 Attractor(b) K 1 = 0, K 2 = 1dφ 1= ω 1 + K 1 sin(φ 2 − φ 1 )dt(5)dφ 2= ω 2 + K 2 sin(φ 1 − φ 2 )dt(6) sin , . , φ = φ 1 − φ 2 . φ (5) (6) :dφdt = ω 1 − ω 2 + K 1 sin(−φ) − K 2 sin(φ) (7)= (ω 1 − ω 2 ) − (K 1 + K 2 ) sin(φ) (8) ω 1 , ω 2 , . 2 . , . () . : , . , . 2(a) , (b) 2 , 0 .3·2 , :1. , 0 . 0 , .2. , . 0 ., . , , . , .28 日 本 ロッ 術 講 (200 年 22 日 〜24 日

RJ200H2Audio signalHumanThereminControls arotation speedCurrent phaseRobot’smotionBeat trackingw hHuman oscillatorA model sendsa commandwhen a phase is zero.Motor commandEnsemble modelInteractionw rRobot oscillatorScoreTheremin controllerFig.3 .✓2 :dφ h= ω h + K h sin(φ r − φ h )dt(9)dφ r= ω r + K r sin(φ h − φ r ).dt(10) ω r :ω r ← ω r + µ(ω r − ω h ), (11), φ r φ h , K r K h, µ .✒3·3 3 ., (1) , (2) , (3) (Ensemble model) . , [3] .. . (1) , . (1) ω h , φ h 0 . ω r , ω h , ω r 0 .4. 3 ,. (1) “” , . (2), . (3) . HRP-2 , ✏Moog Music Etherwave Theremin , 50 cm . “AuraLee” ( 4). , : K r = 0.4, ω h =ω r = 2π/700, µ = 0.01. 50 msec . K h , 1, 2 0, 3 K h = 0.4 ., (60-120 bpm ) , 4 : 66, 80, 100, 112 bpm . 3 , .✑Mean onset error [sec]0.500.450.400.350.300.250.200.150.100.050.00Fig.4 Aura Lee Our methodBaseline methodWorst error66 80 100 112Tempo of metronome [bpm]Fig.5 1: [5] .4·1 error = 1 NN∑mini=1,...,Mj=1∣ onsett (i) − onset d (j) ∣ , (12), N , M , onsetd(j) onset t (i) j i .4·2 1: 5 . , . , , , 1/2 . , . , . , 39% ., , . , 1 msec .4·3 2: , , , 10% .28 日 本 ロッ 術 講 (200 年 22 日 〜24 日

RJ200H2Mean onset error [sec]0.30.250.20.150.10.050Our methodBaseline method66 80 100 112Tempo of metronome [bpm]Fig.6 2: Mean onset error [sec]0.300.250.200.150.100.050.00Our methodBaseline method66 80 100 112Tempo of metronome [bpm]Fig.7 3: 6 . 5 ., 80 bpm , . , µ . µ , . 1 , 120 msec .4·4 3: , . 1, 2 , , . 7 . 5 . , , 14% . 2, . , . , . , 100, 112 bpm 8% , 66, 80 bpm 20% . , .4·5 . 1 , 1 msec . 97 msec , . , , , ., , ,., 2 120 msec . , , 1 120 , . . 2 , . 2, . , , .5. , ., . , 39% . , , π ., , . ,, . , , . . , (No. 19100003, 22118502) GCOE .[1] G. Weinberg et al. The creation of a multi-human, multirobotinteractive jam session. NIME, pp. 70–73, 2009.[2] K. Petersen et al. Development of a aural real-time rhythmicaland harmonic tracking to enable the musical interactionwith the waseda flutist robot. IROS, pp. 2303–2308, 2009.[3] T. Mizumoto et al. Human-robot ensemble between robotthereminist and human percussionist using coupled oscillatormodel. IROS, 2010. to appear.[4] J. Solis et al. Development of Waseda flutist robot WF-4RIV: Implementation of auditory feedback system. ICRA,pp. 3654–3659, 2008.[5] T. Otsuka et al. Music-ensemble robot that is capable ofplaying the theremin while listening to the accompanied music.IEA/AIE, pp. 102–112, 2010.[6] S. H. Strogatz. SYNC: The Emerging Science of SpontaneousOrder. Hyperion, 2003.[7] E. W. Large and M. R. Jones. The dynamics of attending:How people track time-varying events. Psychological Review,Vol. 106, No. 1, pp. 119–159, 1999.[8] R. B. Dannenberg. An on-line algorithm for real-time accompaniment.ICMC, pp. 193–198, 1984.[9] J. Mac Ritchie et al. Visualizing musical structure throughperformance gesture. ISMIR, pp. 237–242, 2009.[10] Y. Kuramoto. Chemical Oscillations, Waves, and Turbulence.Dover Publications, 2003.28 日 本 ロッ 術 講 (200 年 22 日 〜24 日