isiZulu
Ikusasa leMojuli yeKhamera ezimotweni ezizimele: Ubuchwepheshe, Iziqondiso, kanye Nomthelela Oguqulayo
Kwadalwa ngo 10.28
Izimoto ezizimele (AVs) sezingasaphumi emcabangweni wesayensi - seziphakathi kokwamukelwa okujwayelekile, ngokuhamba kancane.amamojula ekhandaukusebenza njenge "amehlo" avumela lezi zimoto ukuba ziqaphele futhi zisebenzisane nezwe. Njengoba ubuchwepheshe be-AV buhamba phambili ukusuka ku-Level 2 (ukuzimela okwengxenye) kuya ku-Level 5 (ukuzimela okuphelele), amamojula wekhamera ayashintsha ngokushesha ukuze ahlangabezane nezidingo zokuphepha, ukunemba, nokwethembeka. Le ndatshana ihlola isimo samanje, ukuphumelela kwezobuchwepheshe, izinselelo, kanye nomkhondo wesikhathi esizayo wamamojula wekhamera ezimotweni ezizimele, ikhanyisa ukuthi azokwakha kanjani isikhathi esisha sokuhamba.
Indima Yamanje Yama-Module Wekhamera Ekuhamba Ngokuzimela
Namuhla, ama-module ekhamera ayisisekelo sohlelo lwezokuxhasa abashayeli (ADAS) kanye nezimoto ezizimele ezisemqoka. Asebenza ngokubambisana ne-LiDAR, i-radar, kanye nezinsiza ze-ultrasonic, athwebula idatha yokubona enezinga eliphezulu ukuze asekele imisebenzi ebalulekile: isixwayiso sokuphuma emgqeni, ukuhamba okuzenzakalelayo kokuphuthumayo, ukulawula ukuhamba okushintshashintshayo, kanye nokutholwa kwabantu. I-AV ejwayelekile ingafakwa ngama-camera angu-8 kuya kwangu-12, abekwe ezungeze imoto ukuze anikeze umbono we-360-degree—kusukela kumakhamera anobubanzi bokubona obuseduze kuya kumakhamera e-telephoto ukuze kutholwe izimpawu zomgwaqo nezithiyo ezikude.
Kwenzenjaniamamojula ekhameraokungasindiwe yikhono labo lokuhumusha umongo wezithombe. Ngokwehlukile ku-radar (okwenza kahle ekukaleni ibanga nokushesha) noma i-LiDAR (okwakha ama-3D point clouds), amakhamera angahlukanisa phakathi komuntu ohamba, umgibeli, kanye nepulasitiki epholile edlula emgwaqweni—kanti futhi akwazi ukuhlonza izibani zomgwaqo, imigoqo, nezimpawu zomgwaqo. Le mbono yomongo ibalulekile ukuze ama-AV enze izinqumo ezisheshayo, eziphephile. Nokho, amamojula amakhamera anjengamanje abhekene nezithiyo: abavumelani nezimo zokukhanya okuphansi, imvula enamandla, noma umoya, futhi ukusebenza kwabo kungaphazamiseka ngenxa yokukhanya okukhanyayo noma uthuli ezibukweni. Lezi zikhala ziqhuba igagasi elilandelayo lokwakha.
Izinguquko Zobuchwepheshe Ezishintsha Imodyuli Zekhamera
Ikusasa lezi zinhlelo zokusebenza zezikhamuzi ku-AVs lichazwa yizinto ezine ezibalulekile zobuchwepheshe, ngayinye ibhekana nezithiyo ezibalulekile futhi ivula amakhono amasha.
1. Izinsiza Zokuhlola Ezine-High-Resolution Ne-Multi-Spectral
Isixazululo asisona nje "izithombe ezicacile"—sikhuluma ngokubamba imininingwane emincane engaba nomthelela phakathi kokuphepha nokubhekana nengcuphe. Amamojula amakhamera ezizukulwane ezizayo asehamba phambili kwezi sensor ze-8MP ukuya ku-12MP, 16MP, futhi ngisho ne-20MP. Ukuxazulula okuphezulu kuvumela ama-AV ukuthi athole izinto ezincane (njengokuphuka emgwaqeni) ukusuka kude, kunika i-AI yemoto isikhathi esithe xaxa sokuphendula. Isibonelo, ikhamera ye-16MP ingakwazi ukuhlonza umgodi emgwaqeni ongamamitha angama-100 phambili, uma kuqhathaniswa namamitha angama-50 nge-sensor ye-8MP—okubalulekile ekushayeleni emgwaqweni ophakeme ngesivinini esikhulu.
Ngaphandle kokukhanya okubonakalayo, amakhamera amaningi-izigaba athola ukuthandwa. Lezi zinsiza ziqoqa idatha ezivela ezingxenyeni ezingabonakali ze-spectra ye-electromagnetic, njenge-near-infrared (NIR) kanye nokuthwebula okushisayo. Amakhamera e-NIR asebenza kahle ezimeni zokukhanya okuphansi, akhipha isidingo sokukhanya okukhulu okukhanyisayo okukhanyisa abanye abashayeli. Amakhamera okushisa, ngakolunye uhlangothi, athola izimpawu zokushisa, okwenza kube lula ukubona abantu noma izilwane ebumnyameni obugcwele noma emoyeni ophakeme—izimo lapho amakhamera okukhanya okubonakalayo kanye ne-LiDAR angase aphumele ngaphandle.
2. Ukuhlanganiswa kwe-AI eMkhawulweni
Inani ledatha elikhiqizwa yizigxivizo ze-AV likhulu: ikhamera eyodwa ye-4K ingakhiqiza i-100GB yedatha ngehora. Ukudlulisa le datha yonke kuseva ye-cloud ephakathi ukuze processing kudala isikhathi sokulinda, okungamukeleki kuma-AV adinga ukuphendula ngemizuzwana. Ukuze kulungiswe lokhu, izigxivizo zezikhamera zifaka ukucubungula kwe-AI "emaphethelweni"—ngqo ngaphakathi kwedivayisi uqobo.
Amachips e-Edge AI, afana ne-NVIDIA's Jetson noma i-Qualcomm's Snapdragon Ride, ayancipha ukuze afaneleke ngaphakathi kumamojula wekhamera. Lawa machips angasebenza kumamodeli okufunda okukh lightweight ukuze ahlukanise, ahlaziye, futhi abeke phambili idatha ngesikhathi sangempela. Isibonelo, esikhundleni sokuthumela wonke umfanekiso wevidiyo kukhompyutha enkulu yemoto, umamojula ungakwazi ukuveza ngokushesha imifanekiso ekhombisa ushintsho oluphuthumayo emgwaqweni ngemoto eseduze, ngenkathi uphonsa phansi imifanekiso engabalulekile (njengomgwaqo ongenalutho). Lokhu kwehlisa isikhathi sokulinda, kwehlisa ukusetshenziswa kwe-bandwidth, futhi kuthuthukisa isikhathi sokuphendula kwemoto.
3. 3D Imaging and Stereo Vision
Ngenkathi amakhamera e-2D ehlisa idatha yokubuka efanele, ukwakhiwa kwe-3D kufaka ubukhulu bokubona—ikhono elibalulekile ukuze ama-AV akwazi ukukala ubude ngok准确. Amamojula amakhamera e-stereo vision, asebenzisa amalensi amabili (njengamehlo abantu) ukuze athathe izithombe ezihlangene, abala ubukhulu ngokukala umehluko phakathi kwemibono emibili. Le teknoloji iya ngokuya iba encane futhi ithengeka, ithatha indawo yezinhlelo ze-LiDAR ezinkulu kwezinye izicelo ze-AV eziphansi (njengama-robot wokulethwa noma izithuthi zesikhungo).
Ngokwezimoto ezisheshayo, amakhamera e-time-of-flight (ToF) avele njengokuguqula umdlalo. Amamojula e-ToF akhipha ukukhanya kwe-infrared futhi akala isikhathi esithathwa ukukhanya ukubuyela emuva ezicini, akha imephu ye-3D enembile yendawo. Ngokwehlukana kokubona, i-ToF isebenza ezimeni zokukhanya okuphansi futhi ingathola izinto ezihambayo ngokunembile kakhulu. Abakhiqizi abambalwa bahlanganisa i-ToF namakhamera e-2D ajwayelekile ukuze bakhe amamojula "ahlanganisiwe" anikeza kokubili umongo (kusuka ku-2D) nokujula (kusuka ku-3D)—ukuhlanganiswa okunamandla kwe-Level 4 ne-5 yokuzimela.
4. Ukuhlala isikhathi eside kanye Nezinhlelo Zokuzihlambulula
Amamojula ekhamera kuma-AV asebenza ezimweni ezinzima: izinga lokushisa eliphezulu (kusuka ku -40°C ebusika kuya ku 85°C ehlobo), imvula, iqhwa, uthuli, kanye nosawoti bezindlela. Ngisho nokungcola okuncane esibukweni kungavala imisebenzi ye-ADAS, kubeka abagibeli engozini. Ukuze kubhekwane nalokhu, abakhiqizi bakha amamojula ekhamera aqinile anemigomo ye-IP69K yokungena kwamanzi nokungena kuthuli. Lezi zinsiza zisebenzisa izinto eziphikisana nokushisa (njenge-ceramic noma ipulasitiki eqinisiwe) kanye nezikhwama ezivaliwe ukuvikela izingxenye zangaphakathi.
I-technology yokuzihlambulula iyinoveli ethola amandla. Ezinye izigaba zifakwe ama-nozzles amancane akhipha umswakama wamanzi (noma isixazululo samanzi-ne-alcohol) phezu kwe-lens, kanti i-micro-wiper ilandela ukususa uthuli. Ezinye zisebenzisa ama-coatings anokuphikisa amanzi aphikisa amanzi nothuli, avimbela ukuhlanganiswa ekuqaleni. Ezindaweni ezibandayo, ama-lens afudumele asusa iqhwa nesnow, aqinisekisa ukubona okungavimbeli unyaka wonke. Lezi zinguquko zokwakha zibalulekile ukuze kuqinisekiswe ukuthi ama-AV reliable kuzo zonke izifunda zomhlaba.
Izinkinga Eziyinhloko Eziphakathi Kwekusasa Lezinhlelo Zekhamera ze-AV
Ngaphandle kwalezi zintuthuko, kunezinkinga eziningi okufanele zixazululwe ngaphambi kokuthi amamojula ekhamera akwazi ngokuphelele ukuvumela ukuzimela kweziqu zezi-5.
1. Ukuqinisekiswa Kwezemvelo
Ngenkathi amakhamera amaningi e-spectral kanye ne-thermal ethuthukisa ukusebenza ezimweni ezinzima, akukho ubuchwepheshe bekhamera obuphephile ngokuphelele. Ukhukhula okukhulu kungafihla ama-lenses, futhi umoya omkhulu ungaphazamisa ukukhanya, kwehlisa ukujula kwesithombe. Ngisho nezinsiza ezinhle kakhulu ziba nezinselelo ngenxa yokukhanya okukhanyayo kwelanga noma izibani ezisondela. Ukuxazulula lokhu kuzodinga hhayi kuphela imishini engcono, kodwa futhi ama-algorithms wesofthiwe athuthukile—njengemodeli ye-AI eqeqeshwe ezinkulungwaneni zezimo zezulu ezinzima—ukuze “gcwalise izikhala” uma idatha yokubona ingaphelele.
2. Ubumfihlo Bedatha Nokuphepha
Amamojula wekhamera abamba inani elikhulu ledatha yokubona, kuhlanganise nezithombe zabahambi, izakhiwo, nezinye izimoto. Lokhu kukhuphula ukukhathazeka ngokuqinisekisa ubumfihlo: le datha igcinwa kanjani, ubani onokufinyelela kuyo, futhi igcinwa isikhathi esingakanani? Ngaphezu kwalokho, amamojula wekhamera ayathinteka ezinsongweni ze-cyber. Abaphangi bangase baphathe idatha yokubona (isb. ukwenza i-AV icabange ukuthi ukukhanya okubomvu kukhanya okukhanyayo) noma baphule umojula ngokuphelele. Abakhiqizi kumele benze ukufakwa kokuvikela kokugcina kokudluliswa kwedatha nokugcina, kanye nezinhlelo eziqinile zokuvikela i-cyber ukuze kuvinjwe ukuhlinzwa.
3. Izindleko kanye Nezinga Lokujwayela
Amamojula amakhamera aphezulu, ahlanganiswe ne-AI, abiza kakhulu—amanani awo manje aphakathi kuka-200 no-500 nganye. Ukuze kube ne-AV enamakhamera angu-12, lokhu kwengeza u-2,400 kuya ku-6,000 enanini lemoto, okuyisithiyo sokwamukelwa okujwayelekile. Njengoba ukukhiqiza kukhula, kulindeleke ukuthi izindleko zehle, kodwa abakhiqizi kufanele futhi balinganisela phakathi kokuthengeka nokusebenza.
Ukulawulwa kwezimiso kuyindaba ehlukile. Azikho izindinganiso zomhlaba wonke eziphathelene nezincazelo ze-module ye-AV camera (isb., isixazululo, indawo yokubuka, amafomethi wedatha). Lokhu kwenza kube nzima ukuthi izingxenye ezahlukene ze-AV (amakhamera, i-LiDAR, amakompyutha amaphakathi) zisebenze ndawonye ngaphandle kokuphazamiseka, kuholele ekwehliseni ukuvuselelwa. Izinhlangano zezimboni ezifana ne-International Organization for Standardization (ISO) zisebenza ekuthuthukiseni izindinganiso, kodwa inqubekela phambili ihamba kancane.
Izitayela Zesikhathi Esizayo: Okufanele Ube Nakho Ngaphambi Kwe-2030
Bheka phambili eminyakeni ezayo, izitayela ezintathu zizobusa ukuthuthukiswa kwemamojula yekhamera ezimotweni ezizimele.
1. Ukuhlanganiswa neLiDAR neRadar
Ikusasa lokubona kwe-AV alikho “ikhamera vs. LiDAR” kodwa “ikhamera + LiDAR + radar.” Amamojula ekhamera azohlanganiswa ngokwanda nezinye izinzwa ukuze kwakhiwe uhlelo “lokuhlanganiswa kwezinsiza” oluzokhokhela ubuthakathaka obuthile. Isibonelo, i-LiDAR inikeza idatha ethile yokujula emoyeni, kanti amakhamera engeza ukuqonda kwesimo; i-radar ibona isivinini nokude emvula enkulu, kanti amakhamera abona uhlobo lwezinto. Lokhu kuhlanganiswa kuzokwenziwa ngezifomethi zedatha ezijwayelekile kanye nezikhompuyutha ezinamandla ezizokwazi ukuhlanganisa idatha evela emithonjeni eminingi ngesikhathi sangempela.
2. Ukunciphisa Usayizi Nokuhlanganiswa
Njengoba ubuchwepheshe buqhubeka phambili, ama-module wekhamera azoba mancane futhi ahlanganiswe kakhulu ekwakhiweni kwemoto. Esikhundleni sama-khamera amakhulu afakwe ophahleni noma ezithombeni zokuqapha, ama-module azofakwa efasiteleni, emgqeni, noma ngisho nasezibani. Ukuze kube mancane kuzovumela ukuthi kwengezwe ama-khamera amaningi—amanye ama-AV angase abe nama-khamera angama-20 noma ngaphezulu ukuze kube nokubona okunembile kakhulu. Ngaphezu kwalokho, ama-module wekhamera azohlanganiswa nezinye izisebenzi, njengezibani ze-LED noma izinhlelo zokuxhumana, kunciphisa isisindo nezindleko.
3. Ukuqhubeka nokuphila kanye Nokwakhiwa Okuphindaphindiwe
Imboni yezimoto iyashintsha iye ekugcineni, futhi amamojula wekhamera awawona umphumela. Abakhiqizi bazosebenzisa izinto eziphinde zisetshenziswe (njengokusebenzisa ipulasitiki ephinde yasetshenziswe ukuze kube nezikhala) futhi bakhe amamojula ukuze kube lula ukuwaphola nokuwaphinda. I-Edge AI izodlala indima ekugcineni: ngokunciphisa ukudluliswa kwedatha kuya efwini, amamojula wekhamera azokwehlisa ukusetshenziswa kwamandla kwemoto. Ezinye izinkampani zifuna ngisho namamojula wekhamera asebenza ngelanga, asebenzisa amapaneli elanga amancane ukuze anikeze amandla kumasensori aphansi, okwenza kube nokunciphisa kakhulu umthelela wemoto emkhathini.
Isiphetho
Amamojula wekhamera angamaqhawe angaziwa kwezobuchwepheshe bezimoto ezizimele, futhi ukuthuthuka kwawo kuzoba nendima ebalulekile ekwamukelweni kabanzi kwezimoto ezizimele. Kusukela kumasensori anokuhlola okuphezulu kanye ne-AI ye-edge kuya ekwenzeni izithombe ze-3D nasekuthuthukiseni okwakhiwe kokuzihlambulula, ukuphumelela kwezobuchwepheshe kuhlangabezana nezithiyo zamanje futhi kuvula amakhono amasha. Ngenkathi izinselelo ezifana nokwethembeka kwemvelo, ubumfihlo, nezindleko zihlala, ikusasa likhanya: ngo-2030, amamojula wekhamera azoba mancane, aqonde kahle, futhi aphumelelayo, asebenza ngokuhlanganyela namanye amasensori ukuze akhe izimoto ezizimele eziphephile, ezithembekile, nezitholakalayo.
Njengoba "amehlo" e-AVs, amamojula wekhamera awawona nje ama-component—ngokuyisisekelo, ayisisekelo sokuguqulwa kokuhamba. Kubakhiqizi bezimoto, izinkampani zobuchwepheshe, kanye nabathengi, ukuqonda ikusasa labo kubalulekile ekuphatheni indlela ezayo.
Uxhumane
Sicela uxhumane nathi uhambele
WhatsApp
WeChat