isiZulu
Amakhamera e-AI Okulandelela Izitayela Zokuhamba Kwabahamba Ngezinyawo: Kusuka Edateni Kuya Ezingqimweni Zedolobha Ezizinzile
Kwadalwa ngo 01.31
Imizi emhlabeni wonke yenza ushintsho olukhulu olubheke ekusimameni, lapho ukuhlelwa kwamadolobha okugxile ezihambini kuvela njengomgogodla walokhu kuguquka. Njengoba omkhandlu omningi uzabalaza ukunciphisa ukukhishwa kwekhabhoni, ukuthuthukisa ukuphepha komphakathi, nokuthuthukisa izinga lempilo, ukuqonda izindlela zokuhamba zezihambeli kube kubalulekile. Izindlela zakudala zokulandelela ukunyakaza kwezihambeli—njengezinhlolovo ezenziwa ngesandla kanye nedatha yenzwa elinganiselwe—asezakwanele ukuhlangabezana nezidingo zezindawo eziguquguqukayo zasemadolobheni. Ngena Amakhamera anikwe amandla yi-AI: amathuluzi athuthukisiwe ashintsha indlela esiqoqa ngayo, sihlaziye, futhi senze ngayo idatha yokuhamba kwabantu. Kulesi sihloko, sihlole ukuthi amakhamera e-AI achaza kanjani ukulandelela izitayela zabahamba ngezinyawo, izinhlelo zawo ezintsha emadolobheni omhlaba, ukuqhekeka kobuchwepheshe okwenza aphumelele kakhulu, kanye nezinhlaka zokuziphatha eziqinisekisa ukusetshenziswa ngokuzwela.
Ngaphezu Kokubala: Ukuthuthukiswa Kokulandelelwa Kwabantu Okusekelwe ku-AI
Sekuphelile izikhathi lapho ukubuka abahamba ngezinyawo kwakugcina nje ekubaleni izinhloko. Amakhamera e-AI amanje ahlanganisa umbono wekhompyutha, izindlela zokufunda komshini, kanye ne-edge computing ukuletha imininingwane ebalulekile ekuziphatheni kwabahamba ngezinyawo. Ngokungafani nezinhlelo zokugada ezijwayelekile, lezi zixazululo eziholwa yi-AI zingakwazi ukukhomba hhayi nje inani labahamba ngezinyawo, kodwa futhi nezindlela zabo, isivinini, ukuziphatha kwamaqembu, ngisho nenhloso—ukuhlukanisa phakathi kwabantu abahamba ngokushesha baye emsebenzini, izivakashi ezihlola indawo, kanye nemindeni ehamba kancane epaki.
Inhliziyo yaleli khono kukhona ama-algorithm athuthukile njenge-YOLOv7 (You Only Look Once) yokuthola izinto ngesikhathi sangempela kanye ne-SORT (Simple Online and Realtime Tracking) yokugcina ukuhlonza okungaguquki kwabantu kuwo wonke amafreyimu ekhamera. Lezi zibuchwepheshe zisebenza ngokubambisana ukucubungula ividiyo endaweni (nge-edge computing) noma efwini, kudala i-metadata esebenzisekayo ngaphandle kwesidingo sokugcina izithombe ezinkulu ezingavuthiwe. Lokhu kusebenza kahle kuyashintsha imidlalo emadolobheni, kuvumela ukuqapha okuqhubekayo ngaphandle kokudlula kakhulu izinsiza zokugcina idatha noma i-bandwidth.
Umehluko obalulekile phakathi kwamakhamera e-AI nezindlela zokulandelela ezijwayelekile yikhono lawo lokuzivumelanisa nezindawo eziyinkimbinkimbi. Angasebenza kahle ezindaweni ezinokukhanya okuphansi, izixuku eziningi, kanye nesimo sezulu esiguquguqukayo—izinselelo ezivame ukuvimbela izinhlelo ezisekelwe ezintweni ezingokwemvelo noma izinzwa eziyisisekelo. Ngokwesibonelo, inzwa yesithombe ye-Sony IMX500, efakwe ukucubungula kwe-AI okusezingeni eliphezulu, ingathola abahamba ngezinyawo futhi ikhiphe idatha yokuhamba ngesikhathi sangempela ngenkathi isebenza ngokuzimela emithonjeni yamandla yangaphandle, okwenza ibe yinhle ezindaweni zasemadolobheni ezikude noma ezinzima ukufinyelela kuzo.
Izibonelo Zamazwe Ngamazwe: Amakhamera e-AI Aklama Ukuhamba Emadolobheni
Amadolobha emhlabeni wonke asevele asebenzisa amakhamera e-AI ukuguqula ukuhlelwa kokuhamba kwabantu. Kusukela ekwenzeni izindawo zomphakathi zibe ngcono kuya ekuthuthukiseni ukuphepha kwethrafikhi, lezi zicelo zangempela zibonisa inani elisebenzayo lobuchwepheshe.
1. Milan: Ukuphinda Ucabange Izindawo Zomphakathi Ngokuqonda Ukuziphatha Kwabantu
E-Piazza Duomo edumile yaseMilan, abacwaningi basebenzise ukuhlaziywa kwevidiyo ye-AI ukuhlaziya izindlela zabahamba ngezinyawo futhi banikeze imininingwane kumaphrojekthi okuvuselela izindawo zomphakathi. Ithimba lafaka amakhamera e-AI anezinhlobo zokuthola izinto nokulandelela ukuze kuthunjwe idatha ngobuningi babahamba ngezinyawo, isivinini, nezindlela zabo. Ngokuhlukanisa abahamba ngezinyawo kumaphrofayela ahlukene—abahambi njalo, izivakashi, abantu abahamba bodwa, namaqembu—idolobha lazuza imininingwane esekelwe ebufakazini ukuze liqanjwe kabusha indawo ukuze kube nokuhamba okungcono nokusebenziseka kalula. Umphumela? Indawo enobungane kubahamba ngezinyawo evumelana nezidingo ezahlukene ngenkathi igcina isimo sayo esibalulekile. Le ndlela igcizelela ukuthi idatha yokuhamba etholwe yi-AI ingasusa kanjani ukuhlelwa kwamadolobha ngaphezu kwemibono ukuya ekwenzeni izinqumo ezisekelwe kudatha.
2. New York & Florida: Ukunciphisa Ukufa Emgwaqweni Ngokuphepha Kwabahamba Ngezinyawo nge-AI
Inisiativu i-Vision Zero, ehlose ukuqeda ukufa kwethrafikhi, ithole umngane onamandla emakhamereni e-AI. Izinkampani ezifana ne-Stop for Kids zisebenzise izinhlelo eziholwa yi-AI emaphuzwini aseNew York naseFlorida ukuze ziqaphe ukuziphatha kwabashayeli futhi ziqinisekise imithetho yethrafikhi evikela abahamba ngezinyawo. Amakhamera la akwazi ukubona "ukuyeka okuyindilinga," ukugijima, nokwehluleka ukunikeza indlela—ukwephulwa okubangela ingxenye yazo zonke izingozi zethrafikhi. Ubuchwepheshe busebenzisa i-AI ukudweba amabhokisi edijithali ezungeze izimoto, buqaphe ukunyakaza kwazo ngaphandle kokuthwebula ubuso noma imisebenzi engaphakathi emotweni. Kuhlolo lwezinsuku ezingu-90 e-Saddle Rock e-Long Island, ukuthobela izimpawu zokuyeka kwanda kusuka ku-3% kuphela kwaba ngu-84%, futhi namuhla kungu-94%. Ngokuvimbela ukushayela okuyingozi, la makhamera e-AI enza izindawo zokudabula ziphephe kakhulu kubahamba ngezinyawo ngenkathi kuqhubekiselwa phambili izinhloso ze-Vision Zero.
3. Roma: Ukulinganisa Ukusebenza Kokuhamba Nokuvikelwa Kobumfihlo
Ukubambisana kweRoma neSony neEnvision kubonisa ukuthi amakhamera e-AI angathuthukisa kanjani ukuhamba ngenkathi kuqala ubumfihlo. Idolobha lafaka izinzwa ze-AI (ezinikwe amandla yi-IMX500 yeSony) ukuze kuqashwe izindawo zokuwela abahamba ngezinyawo, ukuhamba kwabagibeli bamabhasi, nezindawo zokupaka ezitholakalayo. Okusha okubalulekile lapha ikhono lezinzwa lokucubungula idatha endaweni, kuthunyelwa imininingwane yesikhathi sangempela (isb., ukuba khona komhambi ngezinyawo, ukutholakala kokupaka) ngaphandle kokugcina izithombe noma ukudlulisa idatha yomuntu siqu. Le ndlela "yobumfihlo ngokuklama" iqinisekisa ukuthobela imithetho eqinile ngenkathi iletha izinzuzo ezibonakalayo: ukunciphisa ukuminyana kwezimoto ezizungezayo, ukuthuthukisa imizila yamabhasi ngokusekelwe ekubalweni kwabagibeli, nokuthuthukisa ukuphepha kwabahamba ngezinyawo ngokukhanyisa okunokuhlakanipha kwezixwayiso kubashayeli. Kubahambi ngezinyawo, lokhu kusho izikhathi ezimfushane zokulinda amabhasi nokudlula ngokuphepha—konke ngaphandle kokudicilela phansi ubumfihlo.
4. San Jose: Ukukhusela Abahamba Ngezinyawo Abasengcupheni Ngokutholwa Ebusuku
ISan Jose, eCalifornia, isebenzisa izinzwa zombono we-AI zeSony ukubhekana nenkinga ebalulekile: ukushona kwabahamba ngezinyawo phakathi kwabantu abangenamakhaya ebusuku. Idolobha lifake amakhamera e-AI anikwa amandla ilanga ezindaweni ezibalulekile ukuthola abahamba ngezinyawo ezimeni zokungaboni kahle. Noma isesigabeni sokuhlola, ubuchwepheshe buse buhlinzeke ngedatha ebalulekile ngevolumu yethrafikhi kanye nesivinini sezimoto, okusetshenziswa yidolobha ukukhomba izindawo ezidinga ukukhanyiswa okwengeziwe emigwaqweni, izindawo zokuwela, noma izimpawu. Lolu hlelo oluqondile lubonisa ukuthi amakhamera e-AI angalungiswa kanjani ukuze avikele abantu abasengcupheni, kuqinisekiswe ukuthi ukuhlelwa kokuhamba kwamadolobha kubandakanya bonke.
Ukuqhekeka Kobuchwepheshe: Yini Eyenza Amakhamera E-AI Anganamuhla Aphumelele Kangaka?
Impumelelo yamakhamera e-AI ekulandeleleni ukuhamba kwabantu ivela kwezinye izinto ezibalulekile zobuchwepheshe ezixazulula izinkinga zangaphambili:
1. I-Edge Computing Yokuthola Imininingwane Ngokweqile
Izinhlelo zangaphambili zokulandelelwa kwe-AI bezisebenzisa ukubala kwamafu, okwakudala ukubambezeleka nemikhawulo ye-bandwidth. Amakhamera e-AI amanje asebenzisa ukubala komphetho, aqhubekisa idatha ngqo kudivayisi. Lokhu kuvumela ukuhlaziywa kwesikhathi sangempela—okubalulekile ezinhlelweni ezifana nezixwayiso zokuphepha kwethrafikhi nezimpawu zokudlula eziguquguqukayo. Ukubala komphetho kunciphisa nezidingo zokudlulisa idatha, kunciphisa izindleko futhi kuthuthukisa ubumfihlo ngokugcina ulwazi olubucayi lusekhaya.
2. Ukuqashelwa Kwabantu Okunembayo Kakhulu
Izinhlelo ze-Cross-camera pedestrian re-identification (Re-ID) zithuthuke kakhulu, ngezinga lokunemba elifinyelela ku-93.8% ezindaweni ezingavinjelwe kanye no-75.6% ngisho nasezindaweni ezivinjelwe. Lezi zinhlelo zingalandela abahamba ngezinyawo kuwo wonke amakhamera amaningi, zakhe kabusha imizila yazo ephelele edolobheni. Kubahleli bedolobha, lokhu kusho umbono ophelele wokuthi abahamba ngezinyawo bahamba kanjani phakathi kwezindawo zokuhlala, izikhungo zezokuthutha zomphakathi, nezindawo zokuhweba.
3. Ukuhlaziywa Kwabantu Okuhlosiwe
Amamodeli e-AI athuthukile manje angakwazi ukukhomba amaqembu athile abahamba ngezinyawo, njengabantwana, abazali, noma asebekhulile, ngokuhlaziya ubukhulu bomzimba kanye namaphethini okuhamba. Ezindaweni zokuzijabulisa zabantwana, isibonelo, leli khono lisetshenziswe ukutadisha ukusebenzisana phakathi kwabantwana nabantu abadala, okwazisa ukwakhiwa kwezindawo zokudlala ezivikeleke kakhudlwana, ezibandakanya kakhudlwana. Leli zinga lokucaciswa livumela amadolobha ukuthi enze izixazululo zokuhamba ezihambisana nezidingo ezihlukile zabantu abahlukene.
Izinto Ezithinta Ukuziphatha Nokuthobela Imithetho
Njengazo zonke ubuchwepheshe bokuhlola, amakhamera e-AI okulandela abahamba ngezinyawo abeka izinto ezibalulekile ezithinta ukuziphatha kanye nobumfihlo. Ukuqinisekisa ukusetshenziswa okunomthwalo wemfanelo, amadolobha kanye nabahlinzeki bezobuchwepheshe kufanele banamathele ezinhlelweni eziqinile zomthetho kanye nezimiso zokuziphatha.
I-General Data Protection Regulation (GDPR) ye-EU ibeka umthetho womhlaba wonke wokuvikela idatha, idinga "ukunciphisa idatha" (ukuqoqa kuphela idatha edingekayo) kanye nokuba sobala (ukwazisa abantu ukuthi idatha yabo isetshenziswa kanjani). Ubuchwepheshe obufana nezinzwa ze-AI zaseRome buvumelana nale mithetho ngokucubungula idatha endaweni futhi kusulwe ulwazi olungekho ngqo ngesikhathi sangempela. Ngokufanayo, izinhlelo ze-Stop for Kids zicisha izici zobuso nemisebenzi ngaphakathi emotweni, zigxile kuphela ekuhambeni kwezimoto namalayisense ukuze kuqinisekiswe.
Esinye isici esibalulekile ukugwema "ukukhula komsebenzi"—ukwanda kokusetshenziswa kwamakhamera e-AI ngaphezu komgomo wawo wokuqala. Izinhlaka ezibhekayo, njenge-NYCLU yaseNew York, zixwayisa ngokuthi izinhlelo ezisetshenziswa ekuqaleni ukuze kuthuthukiswe ukuphepha kwezimoto zingasetshenziswa kabusha ukuze kuhlale kuhloliwe ngaphandle kwezivikelo ezifanele. Ukunciphisa le ngozi, amadolobha kufanele abeke imithetho ecacile echaza izikhathi zokugcinwa kwedatha, izilawuli zokufinyelela, kanye nokusetshenziswa okwamukelekayo kwedatha yokuhamba etholwe nge-AI.
Ubulungiswa buyabaluleka futhi. Izindlela ze-AI kufanele zihlolwe ukuthola ukuthambekela ukuze kugwenywe ukuthinta imiphakathi esengcupheni ngokungalingani. Ngokwesibonelo, amakhamera kufanele asebenze ngokulinganayo ezimweni ezahlukene zokukhanya futhi aqaphe abahamba ngezinyawo bayo yonke iminyaka, ubulili, nezinhlanga. Ukuhlolwa njalo nokuhlolwa ngabantu besithathu kungasiza ukuqinisekisa ubulungiswa bezindlela.
Ikusasa Lokulandelela Ukuhamba Kwabantu Okunikezwe yi-AI
Sibheke phambili, amakhamera e-AI azoba nendima ebaluleke kakhulu ekwakheni imizi esimeme, enobungani nabahamba ngezinyawo. Izinto eziningana zilungiselelwe ukubumba ukuthuthukiswa kobuchwepheshe:
• Ukuhlanganiswa nezimiso ze-Smart City: Idatha ye-AI yokuhamba izohlangana nedatha esuka ezinhlelweni zezokuthutha zomphakathi, izinhlelo zokuqashwa kwamabhayisikili, kanye nezinzwa zezulu ukuze kwakhiwe izinkundla ezibanzi zokuhamba emadolobheni. Lokhu kuhlangana kuzovumela ukulungiswa okunamandla, njengokuhlela okubukhoma kwezokuthutha zomphakathi ngokwesidingo sabahamba ngezinyawo noma izimpawu zomgwaqo ezivumelanayo ezibeka phambili abahamba ngezinyawo phakathi namahora aphezulu.
• Izindlela Zokuphepha Ezithuthukisiwe: Ukuqhubeka kokufunda okuhlanganisiwe (federated learning) kanye ne-homomorphic encryption kuzovumela amakhamera e-AI ukuthi ahlaziye idatha ngokubambisana ngaphandle kokwabelana ngolwazi oluqondile, kuqhakambise ukuvikelwa kobumfihlo.
• Ukuhlelwa Okuyisinyathelo Sokuhamba: Amamodeli e-AI abikezelayo azosebenzisa idatha yezihambi ezedlule ukubikezela izitayela zesikhathi esizayo, asize amadolobha ukuthi aqagulele izindawo eziminyene futhi ahlele izixazululo ngesinyathelo. Ngokwesibonelo, idolobha lingasebenzisa ukuhlaziywa okubikezelayo ukuhlela izindawo zokuwela ezintsha noma ukwandisa imigwaqo yabahamba ngezinyawo ngaphambi kokuthi kuqedwe ukwakhiwa kwezindlu ezintsha.
• Izicelo Ezihambisana Nesimo Sezulu: Ngokukhuthaza ukusetshenziswa kwezihambi kanye nezokuthutha zomphakathi, idatha yokuhamba etholwe nge-AI izobambisana ekwehliseni ukukhishwa kwe-carbon emadolobheni. Amadolobha angasebenzisa le datha ukukhomba izindawo lapho ukuthuthukisa ingqalasizinda yabahamba ngezinyawo kuzoba nomthelela omkhulu ekwehliseni ukuncika ezimotweni.
Isiphetho: Ukuhamba Okuholwa Yidatha Kwabahamba Ngezinyawo Emadolobheni Angcono
Amakhamera e-AI awasasebenzi kuphela njengamathuluzi okuhlola—sewaba amathuluzi abalulekile okwakha izixeko ezinzile, eziphephile, nezihlanganisayo. Ngokuhlinzeka ngemininingwane ecacile, ngesikhathi sangempela mayelana nezitayela zokuhamba kwabantu, lezi zindlela zobuchwepheshe zenza kube lula kubahleli bezixeko ukwenza izinqumo ezihlakaniphile ezibeka abantu phambili kunezimoto. Kusukela ezindaweni ezihlelwe kabusha eMilan kuya ezindaweni zokuhamba eziphephile eNew York, izifundo zomhlaba wonke zikhombisa izinzuzo ezithile zokulandela abantu nge-AI.
Nokho, amandla aphelele wale teknoloji angafezwa kuphela ngokusetshenziswa okuqondile. Ngokulandela imithetho yokuvikela ubumfihlo, ukuxazulula ukungalingani kwe-algorithm, nokugcina ukuveza, izixeko zingasebenzisa amakhamera e-AI ukudala izixazululo zokuhamba ezizuzisa bonke abantu. Njengoba sibheka ikusasa, ukuhlanganiswa kwe-AI nezinhlelo zokuhamba ezisemadolobheni kuzokuqhubeka nokuchaza ukuthi kusho ukuthini ukuphila edolobheni elinobungane kubantu—elilapho ukuhamba kukhulu, kuphephile, futhi kuhlangene.
Emazibeni aqala lolu hambo, ukhiye ukuthi aqale ngezinhloso ezicacile, abandakanye ababambiqhaza, futhi abeke phambili izinto ezithinta ukuziphatha kuyo yonke imikhakha. Ngendlela efanele, amakhamera e-AI okulandelela ukuhamba kwabantu azoba yisisekelo samadolobha azinzile kusasa.
Uxhumane
Sicela uxhumane nathi uhambele
WhatsApp
WeChat