isiZulu
Ukuvimbela Ukungqubuzana Okusekelwe Embonweni Kwamarobhothi Embonini: Inguquko Yokuhlanganiswa Kwezindlela Eziningi
Kwadalwa ngo 01.28
Imakethe yomhlaba wonke yokuzenzakalela ezimbonini kulindeleke ukuthi ifinyelele ku-$306.2 billion ngo-2027, lapho izinjini zezimboni zizoba nengxenye enkulu yalokhu kukhula. Njengoba izimboni zamukela ama-cobot (ama-robot ahlanganyela) nama-robot ahambayo azimele (ama-AMR) ukuze kukhuliswe ukusebenza kahle, ingozi yokushayisana—phakathi kwama-robot nabantu, ama-robot nemishini, noma ama-robot nezinto ezisetshenzwa—sekuyisithiyo esibalulekile ekuhlanganisweni kwawo okungenazinkinga. Izinhlelo zokugwema ukushayisana ezivamile, ezisekelwe kudatha yenzwa eyodwa noma izindlela ezihlelwe kusengaphambili, zihluleka ezindaweni zezimboni eziguquguqukayo lapho izakhiwo zishintsha, izinto zihamba, futhi abasebenzi babambisana nemishini. Kulapho khona ukugwema ukushayisana okusekelwe embonweni, okwenziwa yiubuchwepheshe bokuhlanganisa izindlela eziningi, iyavela njengento eshintsha imidlalo. Ngokungafani nezixazululo ezijwayelekile, izinhlelo zesimanje ezisekelwe embonweni zisebenzisa ubudlelwano bamakhamera we-2D, i-3D LiDAR, izithombe ezishisayo, kanye ne-edge AI ukuze ziqonde izindawo eziyinkimbinkimbi ngesikhathi sangempela, okwenza amarobhothi enze izinqumo zokugwema ezihlakaniphile, ezivumelana nezimo. Kulesi sihloko, sizohlola ukuthi le ngunguquko yezindlela eziningi isishintsha kanjani ukuphepha kwefektri, izimpumelelo zobuchwepheshe ezenza kube nokwenzeka, imininingwane yokusebenzisa emhlabeni wangempela, nokuthi kungani isibe utshalomali olungadingi ukubuyekezwa kubakhiqizi abacabanga phambili.
Kungani Ukuvimbela Izingozi Kwakudala Kunganele Ezimbonini Zanamuhla
Ngaphambi kokungena ekuqanjweni okusha kwezinhlelo zokubona ezihlangene, kubalulekile ukuqonda imikhawulo yezobuchwepheshe bokuvimbela izingozi ezindala. Sekuyiminyaka eminingi, izimboni ziye zathembele ezindleleni ezimbili eziyinhloko: ukuhlela indlela emisiwe kanye nokutholwa kwenzwa eyodwa.
Ukuhlela ngendlela emisiwe, indlela eyisisekelo kakhulu, ihlanganisa ukuchaza kusengaphambili indlela yokuhamba ye-robhothi endaweni elawulwayo. Nakuba kulula ukuyisebenzisa, le ndlela ayiguquguquki. Uma umsebenzi womuntu, inqola yamathuluzi, noma isithiyo esingalindelekile singena endleleni ehleliwe, i-robhothi ayikwazi ukuyibona—lokhu kuholela ezingozini, ukumiswa kokukhiqiza, noma ngisho nezigameko zokuphepha. Lokhu kungabi bikho kokuguquguquka akuhambisani nezindlela zesimanje "zokukhiqiza eziguquguqukayo," lapho imigqa yokukhiqiza ivame ukushintsha phakathi kwemikhiqizo futhi izakhiwo zasembonini zilungiswa kabusha ukuze kuhlangatshezwane nezidingo eziguqukayo.
Izinhlelo ezinenzwa eyodwa, njengezinzwa ze-ultrasonic noma amakhamera ayisisekelo we-2D, zimelela isinyathelo esiya phambili kodwa zisenezinkinga ezinkulu. Izinzwa ze-ultrasonic zibhekana nezinzwa ezibonakalayo (ezivamile ezimbonini ezinamathuluzi ensimbi) futhi zinokubanga okulinganiselwe, kanti amakhamera we-2D awakwazi ukuthwebula ulwazi olujulile—kwenza kube yinto engenakwenzeka ukukala ngokunembile ibanga phakathi kwe-robot nomgoqo. Ngisho nezinhlelo zokuqala ezisekelwe emehlweni ezisebenzisa i-3D LiDAR kuphela zingavinjwa izimo zokukhanya okuphansi, uthuli, noma ukukhanya okukhanyayo, okusabalele ezimbonini zezimoto, ezikagesi, nezokucubungula ukudla. Lezi zinkinga zisho ukuthi izinhlelo zendabuko zivame ukudinga imigoqo ephephile eqinile (njengezindlu) ukuhlukanisa ama-robot kubantu, okwehlisa injongo yokusebenzisana okubambisanayo nokukhawulela ukusetshenziswa kwesikhala esiphansi.
Inkinga eyinhloko ukuthi izimo zezimboni zishintsha futhi azinakho ukuhleleka. Isikhumbuzo esisodwa noma indlela echaziwe ayikwazi ukubala zonke izinto ezihlukahlukene: umsebenzi ogoba ukuze athathe ithuluzi, iphalethi yezinto ezishiywe okwesikhashana phansi, noma ushintsho oluphuthumayo ekukhanyeni okubangelwa iwindi noma isibani esiphezulu. Ukuze kubhekwane nalokhu, ukuvimbela ukuhlangana okusekelwe emibonweni kufanele kudlule kudatha yomthombo owodwa kube nokubona okuphelele kwezimo—futhi lapho ukuhlanganiswa kwemodi eminingi kuza khona.
Ukwenza Okusha: Ukuhlanganiswa KweMiboniso Eminingi Yokuvimbela Ukuhlangana Okushintshashintshayo
Ukuphuma kwezindlela eziningi ezibonakalayo kuhlanganisa idatha evela ezinhlotsheni eziningi zezinzwa ezibonakalayo (kuhlanganise namakhamera angu-2D, i-3D LiDAR, izithombe ezishisayo, namakhamera e-RGB-D) ngokucubungula kwe-AI emaphethelweni ukuze kwakheke ukuqonda okuphelele, kwesikhathi sangempela kwezindawo ezizungezile zerobhothi. Inzuzo enkulu yalo mkhuba ukuthi inzwa ngayinye ihambisana nobuthakathaka bezinye izinzwa: i-3D LiDAR iletha ukuqonda okunembayo kobude, amakhamera angu-2D abamba umbala nokuthungwa (kusiza ukuhlukanisa phakathi komuntu nento engaphili), izithombe ezishisayo zisebenza ezimweni zokukhanya okuphansi noma ezothuli, namakhamera e-RGB-D agcwalisa isikhala phakathi kwedatha engu-2D kanye no-3D. Lapho kuhlanganiswa ngama-algorithms e-AI athuthukisiwe, lezi zinzwa zakha "imodeli yedijithali" yendawo esheshayo yerobhothi—kwenza kungabi nje ukutholwa kokungqubuzana, kodwa nokugwema okubikezelayo.
Indlela i-Multi-Modal Fusion Esisebenza Ngayo Emaphutheni
Inqubo yokuhlanganisa imibono eminingi ukuze kugwenywe ukushayisana ingahlukaniswa izigaba ezine eziyinhloko, zonke ezicubungulwa ngesikhathi sangempela kumadivayisi angaphandle (ukugwema ukubambezeleka okuvela ekubalweni kwamafu):
1. Ukuqoqwa Kwedatha Ye-Sensor: I-robot ifakwe uhla lwama-sensor ahlelwe ukuze ahambisane nemvelo yefektri. Isibonelo, i-robot yokuhlanganisa ezimoto ingasebenzisa i-3D LiDAR ukuze ibone ukujula, ama-camera e-2D ukuze ibone abasebenzi abantu (ngokusebenzisa umbala nokuqonda), kanye nokubona okushisayo ukuze kutholakale izimpawu zokushisa (kuqinisekiswa ukuthi akukho msebenzi ophuthelwa ezindaweni ezimnyama). I-robot yokucubungula ukudla, ngakolunye uhlangothi, ingase ibeke phambili ama-camera e-2D angangeni manzi kanye ne-3D LiDAR evikela uthuli ukuze ibhekane nezimo ezimanzi nezithuli.
2. Ukulungiswa Kwedatha: Idatha ye-sensor engahlanjwa ihlanzwa futhi ibekwa ezingeni elijwayelekile ukuze kususwe umsindo. Isibonelo, idatha ye-3D LiDAR ihlanzwa ukuze kususwe ukufundwa okungamanga okubangelwa yizicucu zothuli, kanti idatha ye-camera ye-2D ilungiswa ukuze ihambisane nezinguquko zokukhanya. Lesi sigaba sibalulekile ekuqinisekiseni ukuhlanganiswa okunembile—“udoti ungene, udoti uphuma” kusebenza lapha.
3. Ukuhlanganiswa ngeziNqubo ze-AI: IziNqubo ezithuthukisiwe zokufunda komshini (njengeziNqubo ze-convolutional neural networks (CNNs) neziNqubo ze-recurrent neural networks (RNNs)) zihlanganisa idatha elungisiweyo ibe imephu eyodwa yemvelo engu-3D. I-AI ayisihlanganisi nje idatha—iyayichaza. Ngokwesibonelo, ingakwazi ukubona umehluko phakathi kwe-pallet emile (akukho sidingo sokuyigwema ngokushesha) nomsebenzi ohambayo (odinga ukulungiswa komzila ngokushesha). Iphinde ibikezele umzila wokuhamba komgoqelo: umsebenzi ohamba eya kwi-robot uzobangela impendulo ehlukile kunesomunye ohamba kude.
4. Ukwenziwa Kwezinqumo Zokugwema Okuguquguqukayo: Ngokusekelwe kumephu yendawo eyinanyathiselwe, uhlelo lokulawula lwerobhothi lulungisa indlela yalo ngesikhathi sangempela. Ngokungafani nezinhlelo zendlela emile, ezivame ukuma ngokuphelele lapho kutholwa isithiyo (kuphazamisa ukukhiqiza), izinhlelo eziningi zokubona zivumela irobhothi ukuthi lithathe isinyathelo esisebenza kahle kakhulu: yehlise ijubane, lihambe ngendlela yesithiyo, noma lime kuphela uma kudingeka. Leli bhalansi phakathi kokuphepha nokukhiqiza ingenye yezinzuzo ezinkulu zabakhiqizi.
Umthelela Wangempela: Izindaba Zokusebenzisa Ubuhlakani Bezindlela Eziningi Zokubona.
Izinzuzo zemibono zobuhlakani bokugwema ukushayisana ngokusekelwe ezindleleni eziningi zokubona ziyavunywa ezindaweni zangempela zefektri kuzo zonke izimboni. Ake sihlole izindaba ezimbili eziveza inani lazo elisebenzayo:
Isifundo Sokuphumelela 1: Indawo Yokuhlanganisa Izimoto (eJalimane)
Umshayi wezimoto ohamba phambili waseJalimane ubhekene nezingozi phakathi kwama-cobot nabasebenzi emgqeni wokuhlanganisa ibhethri lezimoto zikagesi (EV). Lesi sitshalo sasebenzisa izinzwa ze-ultrasonic ngaphambili, kodwa lezi zehlulekile ukuthola abasebenzi abaguqe noma abedlala eduze kwamarobhothi (isikhundla esivamile ekuhlanganiseni ibhethri) futhi zaphazanyiswa yizinto zensimbi zamabhethri e-EV. Le nkampani yafaka uhlelo oluhlanganisa izindlela eziningi zokubona oluhlanganisa i-3D LiDAR, amakhamera e-RGB-D, kanye ne-AI esezingeni eliphezulu.
Imiphumela ibimangalisa: izingozi zokushayisana kwehle ngo-85% ezinyangeni ezintathu zokuqala. Amandla ohlelo lokuhlukanisa phakathi kwabasebenzi nezinto ezingaphili (njengezikhwama zamathuluzi) kunciphise ukumiswa okungadingekile kokukhiqiza ngo-60%, kwandisa ukusebenza komugqa ngo-12%. Ngaphezu kwalokho, isitshalo sikwazile ukususa amanye amagagasi okuphepha azungeze ama-cobot, kwakhulula isikhala esengeziwe esingu-15% semishini eyengeziwe yokukhiqiza.
Isifundo Sokuphumelela 2: Indawo Yokukhiqiza Izinto Zikagesi (eNingizimu Korea)
Umthengisi wezinto zikagesi waseNingizimu Korea ubhekane nezinselelo ngama-AMR athutha izingxenye phakathi kwemigqa yokukhiqiza. Indawo yayinokuhlelwa okuguquguqukayo, ngokuhlelwa kabusha njalo kwamamodeli amasha ama-smartphone, futhi izinhlelo zekhamera ze-2D zakudala zama-AMR zazilwa nezimo zokukhanya okuphansi ezindaweni zokugcina kanye nokukhanya okuvela ezingxenyeni zengilazi zama-smartphone.
Inkampani yamukela uhlelo olunezindlela eziningi nge-3D LiDAR, izithombe ezishisayo, namakhamera angu-2D anokulungiswa kokukhanya okuguquguqukayo. Izithombe ezishisayo zaqinisekisa ukuthi ama-AMR angakwazi ukubona abasebenzi ezindaweni zokugcina ezimnyama, kuyilapho i-3D LiDAR ihlela ngokunembayo ukwakheka okushintshayo. Imiphumela: Izibalo zokushayisana kwe-AMR zehla ngo-90%, futhi isikhathi esidingekayo sokulungisa kabusha imizila ye-AMR yemigqa yokukhiqiza entsha yancishiswa kusuka emahoreni angu-24 kuya emahoreni amabili. Lokhu kuvumelana nezimo kwavumela umkhiqizi ukuthi andise ukukhiqizwa kwamamodeli amasha ama-smartphone ngama-30% ngokushesha kunangaphambili.
Izinto Ezibalulekile Zokusebenzisa Ukuvimbela Ukungqubuzana Okusekelwe Embonweni Ye-Multi-Modal
Ngenkathi izinhlelo zembono ye-multi-modal zinikeza izinzuzo ezinkulu, ukusebenza kahle kudinga ukuhlelwa ngokucophelela. Nansi imicabango emine ebalulekile abakhiqizi okufanele bayicabangele:
1. Ukukhethwa kwezinzwa Okuhambisana Nemvelo
Akukho nhlobo isethi yenzwa ezilingana zonke. Abakhiqizi kufanele bahlole izimo ezithile ezimbonini zabo: Ingabe imvelo inothuli (isibonelo, ukucubungula insimbi), emanzi (isibonelo, ukucubungula ukudla), noma ikhanyisiwe kahle (isibonelo, ukuhlanganisa izinto zikagesimende)? Ingabe kunezindawo eziningi ezibonakalayo? Ingabe abasebenzi basebenzisa izinto zokuzivikela (njengamajazi abonakala kalula) ezingasiza ekutholeni? Ngokwesibonelo, imboni yezingubo enemicu entantayo ingase iphathe i-3D LiDAR evimbela uthuli futhi igweme ukuthwebula ngezithombe ezishisayo (ezingathinteka uthuli lwemichilo), kuyilapho indawo yokugcina izinto ezibandayo izosebenzisa kakhulu ukuthwebula ngezithombe ezishisayo ukuthola abasebenzi ezimweni ezibandayo, ezingakhanyisiwe.
2. Ukucubungula kwe-Edge AI ukuze Kuncishiswe Ukubambezeleka
Ukugwema ukuhlangana kudinga izinqumo zesikhathi sangempela—ukubambezeleka kwezinye izigidi zemizuzwana kungaholela ezinhlekeleleni. Ukubalwa kwefu kusemva kakhulu kulokhu, ngakho abakhiqizi kumele batshale imali kumadivayisi e-edge AI (njenge NVIDIA Jetson noma i-Intel Movidius) athola idatha ye-sensor endaweni kumrobot noma kumakontroller aseduze. I-Edge AI iphinde iqinisekise ubumfihlo bedatha, njengoba idatha yokwakheka kwefektri kanye nedatha yokukhiqiza engadingeki ukuthunyelwa efwini.
3. Ukuhlanganiswa nezinhlelo zeRobot ezikhona
Abakhiqizi abaningi sebekhuluwe neflethi yamabot ehlukene evela kubahlinzeki abahlukene (isb., i-Fanuc, i-KUKA, i-ABB). Uhlelo lokugwema ukuhlangana olusekelwe emibonweni kumele luhambisane nalezi zinhlelo ezikhona. Bheka izixazululo ezine-APIs ezivulekile (Izixhumi Zokuhlela Izinhlelo) ezingahlanganiswa nezinhlelo zokulawula amabot ezaziwa. Lokhu kugwema isidingo sokushintsha amabot abiza kakhulu futhi kuqinisekisa ukushintsha okungcono.
4. Ukuqeqesha Abasebenzi Nezithunywa ZokMaintenance
Ububanzi obusha bungasebenzi ngempela uma ithimba lazi ukuthi lingalusebenzisa kanjani. Abasebenzi badinga ukuqonda ukuthi uhlelo lokubona lusebenza kanjani (isibonelo, ukuthi lungabathola noma ngisho nasezindaweni ezikhanyayo) nokuthi yini okufanele bayenze uma uhlelo lukhombisa isexwayiso. Amathimba okugcinwa kufanele aqeqeshwe ukuthi alungise izinzwa, abuyekeze izindlela ze-AI, futhi axazulule izinkinga ezivamile (njengokungcola kwezinzwa ngenxa yothuli noma umswakama). Ukutshala imali ekuqeqeshweni kunciphisa isikhathi sokungasebenzi futhi kuqinisekisa ukuthi uhlelo lusebenza kahle kakhulu.
Ikusasa Lokugwema Ukuhlangana Okusekelwe Emibonweni: Yini elandelayo?
Njengoba ubuchwepheshe be-AI nezinzwa buqhubeka nokuthuthuka, ukugwema ukungqubuzana okusekelwe emibonweni eminingi kuzoba namandla nakakhulu. Nansi imikhuba emithathu okufanele uyibheke eminyakeni engu-3-5 ezayo:
• Ukulungiswa kwe-AI Model kumadivayisi e-Edge: Amamodeli e-AI esikhathi esizayo azoba mancane futhi asebenze kahle, okuwavumela ukuthi asebenze kumadivayisi e-edge anomthamo ophansi. Lokhu kuzokwenza izinhlelo eziningi zokubona zifinyeleleke kubakhiqizi abancane abangenakho ukuthenga ihadiwe eliphezulu.
• Ukubambisana phakathi kwamarobhothi: Amarobhothi azokwabelana ngedatha yendawo yawo nomunye nomunye nge-5G connectivity, adale "ubuhlakani obuhlangene" obuhlanganisa yonke indawo yasefektri. Ngokwesibonelo, i-AMR ekupheleni kwefektri ingaxwayisa i-cobot kwenye indawo ngomsebenzi osondelayo, okuvumela ukugwema okuhlangene.
• Ukuhlanganiswa nama-Digital Twins: Idatha yobuhlakani bezindlela eziningi zokubona izohlanganiswa nama-digital twins efektri, okuvumela abakhiqizi ukuthi bahlole izimo zokushayisana futhi balungise izindlela zamarobhothi ngaphambi kokuzisebenzisa endaweni yokukhiqiza. Lokhu kuzoqhubeka nokunciphisa isikhathi sokungasebenzi futhi kuthuthukise ukuphepha ngesikhathi sokufakwa kwesistimu.
Kungani Manje Isikhathi Sokutshala imali ku-Multi-Modal Vision-Based Collision Avoidance
Kubakhiqizi abafuna ukuhlala bephikisana nalesi sikhathi seMboni 4.0, ukuvimbela ukuhlangana akusisimo sokuphepha kuphela—kuyinsiza yokukhiqiza. Izinhlelo zendabuko zikhawulela ukukhiqizwa okuguquguqukayo, kanti izixazululo eziningi ezisekelwe emibonweni zinikeza indlela yokulinganisa ukuphepha, ukusebenza kahle, nokuguquguquka. Izinzuzo zicacile: izingozi ezincane, isikhathi sokuphumula esincishisiwe, ukusetshenziswa okuphumelelayo kwesikhala sokusebenza, kanye nekhono lokwandisa ukuzenzakalela ngaphandle kokubeka engozini ukuphepha kwabasebenzi.
Ngaphezu kwalokho, ingcindezi yokulawulwa kokuphepha kwefektri iyanda emhlabeni jikelele. Umthetho weMishini we-European Union (2006/42/EC) kanye ne-U.S. Occupational Safety and Health Administration (OSHA) baphakamisa izimfuneko eziqinile zokuphepha kwamabhothi, okwenza izinhlelo zokuvimbela ukuhlangana ezithuthukisiwe zibe yisidingo sokuhambisana. Ukutshalwa kwezimali manje akusizi kuphela abakhiqizi ukuhlangabezana nalezi zimfuneko kodwa futhi kubabeka endaweni yokuthola inzuzo kumkhuba okhula wokuhlanganiswa kokusebenza.
Isiphetho
Ukugwema ukungqubuzana okusekelwe embonweni kumalobhothi ezimbonini kuthuthukiswa kakhulu, okubangelwa ukuhlanganiswa kwezinzwa eziningi kanye ne-AI esezingeni eliphezulu. Le ndlela entsha inqoba imikhawulo yezinhlelo zakudala ngokunikeza ukuqonda okuphelele, okwangempela kwezindawo ezimbonini eziguquguqukayo—kwenza amalobhothi enze izinqumo zokugwema ezivumela ukuthi avikele abasebenzi ngenkathi kugcinwa ukukhiqiza kuqhubeka kahle. Izifundo zangempela ezivela ekukhiqizeni izimoto nezinto zikagesi zibonisa izinzuzo zayo ezibonakalayo, kusukela ekunciphisweni kokungqubuzana kuya ekuthuthukisweni kokusebenza kahle nokuvumelana nezimo.
Njengoba abakhiqizi bamukela i-Industry 4.0 kanye nokukhiqiza okuguquguqukayo, ukugwema ukungqubuzana okusekelwe embonweni okuningi kuzoba yisisekelo samasu okuphumelelayo okuphatha ngokuzenzakalelayo. Ngokukhetha ngokucophelela izinzwa ezihambisana nemvelo yazo, ukutshala imali ekucubunguleni kwe-AI emaphethelweni, ukuhlanganisa nezinhlelo ezikhona, nokuqeqesha amaqembu abo, abakhiqizi bangavula amandla agcwele aleli zinga lobuchwepheshe. Ikusasa lokuphatha ngokuzenzakalelayo embonini liphephile, liyakwazi ukuzivumelanisa, futhi lisebenza kahle—futhi umbono okuningi uhola indlela.
Uxhumane
Sicela uxhumane nathi uhambele
WhatsApp
WeChat