Internatonal Journal of Mechancal Engneerng and Robotcs Research Vol. 4, No. 3, July 5 Extenson of the Upper Extremty wth Shoulder Movements Nobuak Nakazawa and oshkazu Matsu Dvson of Mechancal Scence and echnology, Faculty of Scence and echnology, Gunma Unversty, Oota, Japan Emal: {n.nakazawa, matsu}@gunma-u.ac.jp human's characterstcs to task plannng for a moble anthropomorphc manpulator. Such a moton composed of the plural exercses, termed the fuson moton, may nclude a clue for the optmal plannng of the Humanod Robot n the complex tasks. However, only few studes concerned wth the fuson motons have so far been made. We nvestgated the human's fuson motons composed of extendng the upper lmb and bendng the trunk of upper body and examned how these exercses were fused. Utlzng the measured data, the smulaton tool whch could be shfted the acton start tme of the two exercses was made for subjectvty evaluatons to clarfy the specal feature of the fuson exercses wthout sense of ncongruty. Abstract In ths paper, we nvestgated characterstcs of human's upper lmb extenson wth shoulder movements when the grasped object was moved from one pont to other pont. In cases where the goal poston s put n a farther place than the length of the extended arm, not only the velocty profle of the hand by extenson exercse but also that of the shoulder by anteflexon exercse are bell-shaped. hese two exercses are smultaneously performed, but the hand velocty becomes a peak earler n comparson wth the shoulder one. Reproducng the measured human data, the exercses of a three-lnk arm wth shoulder or body movements are smulated and the characterstcs of a human-lkeness fuson are examned by shftng the start tme of the shoulder movement. As an evaluaton result, t was found that the relaton between the velocty peak postons of the hand and shoulder remarkably nfluences on the human-lkeness. Index erms human extenson, smulaton arm, psychologcal Recorder evaluaton, Subject Object INRODUCION In recent years, robot s becomng our famlar exstence and robot moton must have been planned so as not to gve us a threat, sense of ncongruty n the acton generatons. Especally, n cases where robot s workng around us and t has somethng nteracton wth us, movement of hand and arm may play an mportant role as a man-machne nterface. he structure of a human upper lmb s smlar to the robot manpulators, and the human schemes of the arm movements are thought to be avalable for the control of a robot manpulator. herefore, many nvestgatons regardng the human arm movements have been done. In order to do the advanced work, the arm movements wth a restrcton to the hand were also nvestgated. suj et al. []-[3] examned the crank rotaton tasks performed by the human and examned how the arm postures were determned by utlzng knematc redundancy. anaka et al. [4]-[6] nvestgated the characterstcs of the hand trajectores n the manpulaton of a holdng non-holonomc car. If the movement range of hand s spread, the antflexon exercse, rotaton of the upper trunk, or the legs exercses are requred. Sato and Sugano [7] carred out the experment n handlng a manpulator based on a gude shaft wth body movements and appled the obtaned Manuscrpt receved July, 4; revsed June, 5. 5 Int. J. Mech. Eng. Rob. Res. do:.878/jmerr.4.3.33-37 33 Fgure. Expermental scene. 6mm mm I. ６ ５ ４ ３ ２ １ Intal Poston Subject Fgure. Dstrbuton of target. II. UPPER LIMB EXENSIONS WIH SHOULDER MOVEMENS A. Experment Frst of all, we measured human s upper lmb extensons wth shoulder movement. In the experment,
Internatonal Journal of Mechancal Engneerng and Robotcs Research Vol. 4, No. 3, July 5 human subjects grasped a cylndrcal plastc object ([mm] 9[mm], 5[g]) and moved t toward a goal poston. he partcpated subjects were drected so as not to stand from the char, therefore, the grasped object was translated to the goal postons by both extenson exercse of the upper lmb and anteflexon exercse of the upper part of the body as shown n Fg.. he trajectores of the shoulder, elbow, and wrst were measured by shootng them wth the dgtal vdeo camera placed at the upper part 7[mm] of the table. Fg. shows the dstrbuton of the goal postons and the posture of the subject n the ntal state of the experment. Before startng each tral, subjects extended the arm straghtly wth the hand located at the pont 3 so as to determne the char poston (Condton I). After that, keepng the posture of the body, the upper lmb was bent and the hand was returned to the ntal poston (Condton II). Each next poston of sx goals and the ntal poston was dstrbuted wth [mm] span. he healthy human subjects aged from to 4 partcpated n ths experment. Y-axs(mm) Y-axs(mm) 5 5 Subject S mm L-Shoulder R-Shoulder -5 5 X-axs(mm) Subject S 6mm L-Shoulder (a) mm R-Shoulder -5 5 X-axs(mm) (b) 6mm Fgure 3. rajectory of upper lmb and shoulders. B. Results Expermental results are shown n Fg. 3. Fg. 3(a) and (b) show the trajectores of the upper lmb and body under the condtons that the span to the goal was set to [mm], 6[mm]. he goal s near the ntal poston (Fg. 3(a)), the movement of the grasped object s carred out by only extendng the upper lmb wthout movng the shoulder. On the other hand, when the movement span s longer than the arm length (Fg. 3(b)), t can be confrmed that the shoulder also moves to the goal poston wth extendng the upper lmb. Especally, the rght shoulder moves to the frontal drecton remarkably compared wth the left one, therefore, the anteflexon exercse s performed wth the rotaton of the trunk wth respect to ts major axs. In order to nvestgate the upper lmb exercse wthout the movement of the shoulder, the rght shoulder poston s set to the orgn of the upper lmb and the hand s re-defned as follows: P P P () where, P [ x y ] (,,,, N) s the measurement pont of the hand ncludng the shoulder movements, and P [ x y ] ndcates the rght shoulder as shown n Fg. 4. Fg. 5 shows the tme trajectores of a tangental velocty concernng wth the hand poston P, the rght R shoulder poston P, and the relatve hand poston P defned n (). he hand and rght shoulder are movng wth the bell-shaped velocty profle. he velocty of the relatve hand poston s also bell-shaped and t can be confrmed that both extendng exercse of upper lmb and anteflexon exercse synchronze each other. However, there s some temporal dfference wth respect to the locaton of the peak value n the velocty profle. Fg. 6 shows the tme when the velocty of the R P rght shoulder poston and relatve hand poston P become a maxmal value, whch s normalzed wth that of the measured hand poston P. he value of P s approxmately. across the subjects, and the velocty of the hand created by expandng the upper lmb has a maxmal value n the same tme as that of the hand ncludng the shoulder movement. Wth reference to the temporal order, the R P values of P reach a peak earler n comparson wth that of regardless of the goal poston. Namely, the hand movement by the upper lmb exercse has a peak earler n comparson wth the shoulder movements. (Left Shoulder) L L L P [ x, y ] Velocty(mm/sec) 8 6 4 Relatve hand pos. O () P [ x, y ] Fgure 4. Frame model. Orgn of arm (Rght Shoulder) [ x P, y ] Rght shoulder.5.5 me(sec) Fgure 5. Velocty profle. 5 Int. J. Mech. Eng. Rob. Res. 34
Internatonal Journal of Mechancal Engneerng and Robotcs Research Vol. 4, No. 3, July 5 n the statc coordnates (x,y) s shown. For the P parameter of, we can see that the body starts movng after the exercse of the upper lmb extenson has been started. Fg. 9 shows the velocty profle of the hand and body postons n the statc coordnates. he data of were prepared for the evaluatons by movng the P startng tme of the body (,,, 5 ). en frame movng s equal to /3 [sec] (one frame s /3[sec]) and maxmal dfference compared wth the orgnal measured data s 5 / 3 [sec]. Fgure 6. Peak poston of tme duraton. III. Moved frame number ˆ y P PSYCHOLOGICAL EVALUAIONS y A. Smulaton From the temporal pont of vew, there are numerous combnatons between the shoulder movement and the exercse of upper lmb extenson. In ths secton, we nvestgated the permsson ranges of the temporal dfference between the startng tmes of the two exercses so as to generate the fuson motons just lke a human wthout the sense of ncongruty. Here, the smulaton tool whch could shft the start tme of the shoulder movement wth respect to that of the arm movement was made n order to dsplay a frame model on a computer CR, and subjects watched and evaluated these motons. As shown n Fg. 7, the length of the shoulder, brachum, and forearm s l, l and l3, respectvely. In the coordnates (x, y) whose orgn s the pont O, the tme trajectory of the hand Pˆ [ xˆ yˆ ] (,,,, N ) s gven by () Pˆ P P P yl yr ]. x x O Fgure 7. Frame model for smulaton. 3 () (Shoulder) Angle(deg.) 9 6 3 ().5.5.5 me(sec) 6 (4) 4 y y+ ( =) x - -4 P ndcates the hand poston n the coordnate (x', y') where the orgn s set to the pont of P. he jont angle, and 3 are measured data generated by the subjects as shown n Fg. 8. Regardng the human motons, the angle 3 s remarkably changed, and the exercse of upper lmb expand s actvely performed untl the hand reaches the goal poston. Moreover, we can see that the anteflexon of the trunk synchronzes wth the extenson exercse of the upper lmb. P s the body poston n the statc coordnates and t s derved by modfyng the measured data P L, P R as shown n Fg. 4 and by movng the tme as. Fg. 8 also shows the example of the tme trajectores n cases where the start tme of the body P s changed(, ). In the lower fgure, the body poston 5 Int. J. Mech. Eng. Rob. Res. Poston(pxel) x L xr [ l l where, P [ x y ] l cos l cos( ) l3 cos( 3 ) (3), l sn l sn( ) l3 sn( 3 ) P 3 l3.5 x+ ( =).5.5 me(sec) Fgure 8. Example of smulaton. Velocty(mm/sec) 6 Relatve hand pos. 4 Arm extenson antflexon Rght shoulder me duratons.5.5 me(sec) Fgure 9. Velocty profle for evaluatons. 35
Internatonal Journal of Mechancal Engneerng and Robotcs Research Vol. 4, No. 3, July 5 B. Evaluaton Human subjects watched the smulatons dsplayed on a computer CR and evaluated ther motons. he ndcated fuson exercses were nspected wth the method conformed to ratng scale method used for the psychologcal evaluatons of robot motons [8]-[]. Here, we used fve adjectves (smooth - awkward, fast - slow, careful - careless, sklled - unsklled, and human lke - un-human lke''). he subjects were nstructed to emulate the ndcated motons on a computer CR by synchronzng t wth the actual exercse of the upper lmb and anteflexon so as to make t easy to feel the sense of ncongruty concernng wth the fuson exercse. Accordng to the subject's requests, they could watch the smulatons repeatedly. Before the experments, the evaluaton sheets were gven to the subjects and they were flled wth ther mpressons by pantng the crcle at any of the seven stage scales about the fve adjectves. Eleven smulaton patterns were prepared and ndcated from 5 to 5 n order. hrty healthy subjects aged from to 39 partcpated n ths experment. C. Results he results of subjectvty evaluatons are shown n Fg.. Each pont s mean value of all subjects, and the vertcal short bars ndcate the standard devaton. In cases where the parameter was a negatve, the body moves before the upper lmb extendng. For the case of a postve, the above temporal order to start movng was replaced. Wth reference to the evaluatons for the human lke motons, when the two exercses started smultaneously ( ), the fuson moton was felt most human lke. In the case of, the mean value was located at the left sde to the center scale (nether-nor), the boundary of the adjectve pars, and the fuson motons were carred out wthout sense of ncongruty even though whchever of the two exercses starts to move prevously. On the other hand, n the case of, the standard devaton was larger compared wth the other crcumstances, so there were some dfferences concernng wth the evaluatons across the subjects and some of them had a feelng of un-human lke. If the absolute value of was more than 3, all subjects felt a sense of ncongruty. In Fg. 9, n order to make a human lke fuson motons, the bell-shaped velocty profle of the body must have reached a peak value by the tme when the arm extendng moton was completed, n cases where the arms exercsed an exhbton prevously ( ). If the body started to move prevously ( ), the exercse of extendng the arm must have started before the movement of the body reached a velocty peak. In the evaluatons regardng the smoothness, speed, and skll, there was a same tendency as that of the human lke. Each adjectve was felt strongly when the parameter was zero, and all feelng boundary were focusng on the crcumstance of. Accordng to the ncrease of the temporal dfference between the two exercses, awkward, slow, and unsklled were strongly felt. he trend regardng the prudence feelng was dffered from other, the subjects tended to feel careful motons as the temporal dfference was longer. smooth fast careful sklled human lke :+5 :+4 :+3 :+ :+ : :-5 :-4 :-3 :- :- very pretty a lttle Nethernor a lttle pretty Fgure. Evaluaton results. IV. CONCLUSIONS very awkward slow careless unsklled un-human lke In ths paper, we nvestgated the human's upper lmb extensons and shoulder movements. Expermental results sad that velocty profle of both upper lmb and shoulder were a bell-shaped and hand velocty became a peak earler n comparson wth the shoulder one. he motons of upper lmb extenson and body movement were smultaneously performed and t could be confrmed that the relatve poston of the velocty peak remarkably nfluenced on a human-lkeness. he obtaned results may mply a clue of the optmal moton plannng for arm manpulator workng around us humans. In cases where the robot needs to extend the arm wth the body movements, the temporal dfference between the startng tmes of the two exercses must be consdered n order to remove a threat and sense of ncongruty. REFERENCES []. suj and K. Itoh, Bomechancal analyss of human arm posture durng crank rotaton tasks, he Japanese Journal of Ergonomcs, vol. 8, no. 3, pp. 49-57, 99. []. suj and K. Itoh, Poston/Force control strateges n human movements durng crank rotaton tasks, he Japanese Journal of Ergonomcs, vol.8, no. 4, pp. 9-8, 99. [3]. suj, Y. Mk, and K. Itoh, force formaton durng constraned arm movements and vrtual trajectory hypothess, ransactons of the Socety of Instrument and Control Engneers, vol. 8, no. 3, pp. 366-373, 99. [4] Y. anaka, et al., Bo-mmetc trajectory generaton of robots based on human arm movements, Journal of the Robotcs Socety of Japan, vol. 8, no. 5, pp. 699-75,. 5 Int. J. Mech. Eng. Rob. Res. 36
Internatonal Journal of Mechancal Engneerng and Robotcs Research Vol. 4, No. 3, July 5 [5]. suj, Y. anaka, P. G. Morasso, V. Sangunet, and M. Kaneko, Bo-mmetc trajectory generaton of robots va artfcal potental feld wth tme base generator, IEEE ransactons on Systems, Man, and Cybernetcs-Part C: Applcatons and Revews, vol. 3, no. 4, pp. 46-439,. [6]. suj, Y. anaka, and M. Kaneko, Bommetc trajectory generaton based on human movements wth a nonholonomc constrant, IEEE ransactons on Systems, Man, and Cybernetcs-Part A: Systems and Humans, vol. 3, no. 6, pp. 773-779,. [7] M. Sato and S. Sugano, Analyss of coordnate moton between upper body and legs for a moble manpulator, Bomechansms, vol. 4, pp. 49-59, 998. [8] S. Shbata and H. Inooka, Psychologcal evaluatons of robot motons, Internatonal Journal of Industral Ergonomcs, vol., pp. 483-494, 998. [9] S. Shbata and H. Inooka, Emotonal evaluatons on robot motons by usng ratng scale method, he Japanese Journal of Ergonomcs, vol. 3, no., pp. 5-59, 995. [] S. Shbata, K. Ooba, and H. Inooka, Emoton evaluaton of human arm moton models, n Proc. nd IEEE Internatonal Workshop on Robot and Human Communcaton, 993, vol., pp. 346-35. Nobuak Nakazawa was born n 969, Japan. He receved the B.S. from oyama Unversty n 993, and M.S and Dr. Eng. degrees from ohoku Unversty, Japan n 995 and 998, respectvely. From 998 to 999, he was a Research Fellow of Japan Socety for the Promoton Scence (JSPS) at ohoku Unversty. From 999 to 6, he was a Research Assocate of Graduate School of Engneerng, Gunma Unversty, Japan. Snce 7, he has been an Assocate Professor of the Department of Producton Scence and echnology, Gunma Unversty, Japan. Hs research nterests nclude ergonomcs, human nterface, and welfare support-equpment. oshkazu Matsu was born n 954, Japan. He receved the B.S. and M.S degrees from Waseda Unversty, Japan n 977 and 979, respectvely, and the Dr. Eng. degree from Waseda Unversty, Japan n 997. From 98 to 994, he was a Research Engneer of oshba Corporaton. From 994 to 996, he was an Assstant Research Engneer of Advanced elecommuncatons Research Insttute Internatonal (AR). From 996 to 998, he was a Research Engneer of oshba Corporaton. From 998 to 6, he has been an Assocate Professor of Graduate School of Engneerng, Gunma Unversty, Japan. Snce 7, he has been an Assocate Professor of the Department of Producton Scence and echnology, Gunma Unversty, Japan. 5 Int. J. Mech. Eng. Rob. Res. 37