Eminyakeni yamuva nje siye sasuka ekuboneni ama-drone njengento elula ithoyizi lobuchwepheshe obuphezulu lokuqopha amavidiyo ukuzenza ithuluzi elibalulekile kwezokuphepha, ukuqapha, ngisho nezingxabano ezihlomile. Futhi, njengoba kuvame ukwenzeka nganoma yiluphi ubuchwepheshe obuqala ngokushesha kangaka, isici sokuphepha nokuvikela asiqhubeki ngesivinini esifanayo ngaso sonke isikhathi.
Kulesi simo, iqembu labacwaningi abavela eNyuvesi yaseCalifornia, e-Irvine, libonise okuthile okuzwakala sengathi yihlaya, kodwa okuphambene nalokho: Kungenzeka ukuheha nokunciphisa ama-drone athile azimele usebenzisa isambulela esinombala othize kuphela.Akukho mishini yokusakaza imisakazo, akukho ukugenca okukude, akukho zikhali eziyinkimbinkimbi. Isambulela "esingavamile" kanye nobunjiniyela obuphambene nendlela la madivayisi abheka ngayo umhlaba.
Kungani ama-drone azimele ebangela ukukhathazeka okwandayo
Ukwanda kwama-drone akugcini nje ngokuzijabulisa. IShayina, iRussia, i-United States, kanye namanye amazwe Bacindezela ngamandla ukuthuthukiswa kwezimoto zasemoyeni ezingenamuntu ezihlonyiswe ngezinhlelo zobuhlakani bokwenziwa ezikwazi ukwenza izinqumo zodwa. Asisakhulumi nje ngamakhamera okundiza abantu abanethonya, kodwa ngamapulatifomu azimele anezinhlelo zokuqapha kanye nokuqapha.
Isibonelo esicacile kakhulu singatholakala empini yase- ERussia nase-Ukrainelapho kusetshenziswe khona zonke izinhlobo zama-drone: amadivayisi e-kamikaze, amayunithi okulandelela, izinhlelo ezixhunywe ezintanjeni ze-fiber optic ukuthuthukisa ukudluliswa kwedatha, njll. Okungeziwe kulokhu yintuthuko ehambisana nesayensi eqanjiwe, njengokuhlolwa kwaseRussia okusolwayo amajuba aphenduka ama-bio-drone, okubonisa izinga abanye abadlali abazimisele ngalo ukuhlola noma iyiphi indlela yezobuchwepheshe ukuze bathole izinzuzo enkundleni yempi.
Inkinga ukuthi lobu buchwepheshe obufanayo, uma budluliselwa emkhakheni wezakhamuzi, buvula umnyango wezimo ezingakhululekile kakhulu: ukuqapha okuqhubekayo kwabantu, ukulandelela okuzenzakalelayo okungagunyaziwe, ukuqoqwa kwedatha ezindaweni ezibucayi, noma ukungenelela nje kubumfihlo bansuku zonke banoma ubani ohlangana nenye yalawa madivayisi ngesikhathi esingafanele.
Ukwamukelwa ngobuningi kwama-drone anemisebenzi yokulandelela ngokuzimela emisebenzini efana nokuphepha kwangasese, ukuqapha umngceleUkulawula isixuku noma ukuqapha ingqalasizinda ebalulekile kwandisa amathuba abo okusebenza… kodwa, ngesikhathi esifanayo, kwandisa izingozi. Hhayi nje ngoba ama-drone angasebenzi kahle, kodwa ngoba angasetshenziswa, akhohliswe, noma asetshenziswe ngokumelene nalabo abawasebenzisayo.
Ngenxa yalesi simo, kuyacaca ukuthi akwanele ukubhala ngemfihlo ukuxhumana noma ukuvikela ama-radio band. Impi yangempela nayo ilwiwa phansi, endleleni ama-algorithms ombono kanye ne-AI ahumusha ngayo lokho abakubonayo ngekhamera ye-drone kanye nendlela abasabela ngayo kumaphethini okubonwayo aklanyelwe ukuwadida.
Ucwaningo lwe-UC Irvine: isambulela esimelene nobuhlakani bokwenziwa
Esikhundleni sokuthuthukisa i-drone entsha ehlaselayo, ithimba le- IYunivesithi yaseCalifornia, e-Irvine (UC Irvine) Wanquma ukubhekana nenkinga ngakolunye uhlangothi: indlela yokuzivikela ngokomzimba ezinhlelweni zokulandelela ezizimele ezisekelwe ku-AI. Umgomo wakhe wawucacile: ukuthuthukisa lokho okubizwa ngokuthi "ukuphepha kwe-cyber ngokomzimba" izinto zansuku zonke nezingabizi kakhulu, ngaphandle kokuthembela ku-hardware ekhethekile noma ulwazi oluthuthukisiwe lokugenca.
Ngesikhathi sophenyo lwabo, ochwepheshe bahlaziye ngokucophelela ukuthi ama-algorithms okulandelela afakwe kuma-drones amaningi ezentengiselwano asebenza kanjani. Ngokukhethekile, bagxile emisebenzini eyaziwa ngokwezentengiselwano ngokuthi I-Active Track, i-Dynamic Track, noma ezinye izinhlelo ezifanayookuvumela i-drone ukuthi ilandele ngokuzenzakalelayo umuntu noma into ngokusebenzisa umbono wekhompyutha, ngaphandle kokuthi umshayeli wendiza alungise indlela yokuqondisa ngaso sonke isikhathi.
Ngemva kokuhlola okuningi, bathole ubuthakathaka obulula ngendlela emangalisayo: amamodeli athile e-drone asebenza kahle Ukulandelela okuqondiwe kusetshenziswa amanethiwekhi e-neural kanye nokuhlaziywa kokunyakaza kwesithombe Zingadideka uma ziboniswa ngephethini ebonakalayo ethile. Le ndlela yaqanjwa ngokuthi I-FlyTrap"isicupho sezindiza", okubhekisele endleleni eheha ngayo i-drone endaweni lapho ingabanjwa khona noma idutshulwe.
Into ehlaba umxhwele kakhulu ngaleli cala ukuthi ukusetshenziswa okungokoqobo kwe-FlyTrap kwenziwa kusetshenziswa into evamile njengesambulela. Ngokunyathela iphethini ebonakalayo efanele ebusweni besambulela Futhi uma uyivula phambi kwe-drone, i-AI yokulandelela ihumusha uchungechunge lwezinguquko esithombeni ezenza ikholwe ukuthi inhloso iyahamba, yize umuntu oyiphethe enganyakazi.
Imiphumela yethulwe engqungqutheleni yamazwe ngamazwe egxile ekuphepheni kwamakhompyutha (NDSS), lapho kwachazwa khona kabanzi ukuthi le ndlela akuyona nje into eyenzeka elabhorethri. Ukuhlolwa kubonise ukuthi ukuhlaselwa ngokomzimba kusebenza ngaphansi kwezimo ezahlukahlukenengezinguquko ekukhanyeni kanye nesimo sezulu, okusisondeza kakhulu esimweni sokusetshenziswa kwangempela ngaphandle kwendawo elawulwayo.
Indlela ukuhlasela kwe-FlyTrap okusebenza ngayo isinyathelo ngesinyathelo (ezingeni lomqondo)
Ukuze siqonde ukuthi kungani isambulela singabeka i-drone ezimele esimweni esinzima, sidinga ukubukeza ukuthi izinhlelo zokulandelela eziqondiwe zisekelwe kanjani ku- umbono wokufakelwaEmpeleni, i-drone iqhubeka nokuthwebula izithombe zendawo eyizungezile futhi, isebenzisa amanethiwekhi ezinzwa kanye nama-algorithms okuhlaziya ukunyakaza, izama ukuthola into okufanele iyilandele ngaphakathi kohlaka ngalunye.
Lezi zindlela zokuqondisa zigxile ku amaphethini okuma, umbala, umehluko, kanye nokunyakaza ukulinganisa ukuthi isihloko sikuphi esithombeni samanje uma kuqhathaniswa nesangaphambilini. Kusukela kulolo lwazi, babala ukuthi isihloko siyasondela, siyahamba, siya eceleni, sicashe ngemuva kwesithiyo, njll. I-drone ibe isilungisa ijubane layo kanye nomzila wayo ukuze igcine isihloko ngaphakathi kohlaka kude kubhekwa "njengokufanele kakhulu".
Ukuhlasela kweFlyTrap kusebenzisa kahle le ndlela yokucabanga. Iphethini yesithombe eyenzelwe isambulela ibangela ukuthi ikhiqize amaphethini lapho idrone ihamba kancane noma ishintsha i-engeli yayo yokubuka. izinguquko ezibonakalayo esikalini kanye nendawo yenhloso okuyinto i-algorithm eyichaza njengokunyakaza okuqhubekayo kude. I-AI ikholelwa ukuthi isihloko siyahamba sibuyela emuva, kanti empeleni simile singanyakazi.
Ngakho-ke yenzani idrone? Ukuze kulungiswe leli banga elicatshangelwayo, nciphisa ibanga kancane kancaneUkuzama "ukusondela" kumuntu ukuze kugcinwe uhlaka ngokwemithetho yangaphakathi yesistimu yokulandelela. Umphumela uba ukuthi indiza ingena endleleni enamandla kakhulu ejikelezayo... ize isondele kakhulu esambuleleni nomuntu osiphethe.
Uma i-drone iseduze kakhulu, umqhubi wohlaselo unesilinganiso esihle kakhulu: bangakwazi ukuyibamba ngenetha, ukuyishaya, ukuyiphazamisa, noma ukuyenza ishayisane ngokumelene nendawo eseduze. Ngokungafani nezinye izindlela zokuzivikela ezihlose kuphela ukwenza i-drone ilahlekelwe yithagethi yayo noma ukuyeka ukulandelela, i-FlyTrap ikuvumela ukuthi uyibeke endaweni efanele yokungathathi hlangothi ngokomzimba.
Amamodeli e-drone athintekile kanye nobubanzi bangempela benkinga
Abacwaningi base-UC Irvine abazange bayeke ekulingiseni; bahlole ubuchwepheshe babo nge ama-drone angempela ezentengiselwanoPhakathi kwamamodeli ahlolwe ngempumelelo kukhona:
- I-DJI Mini 4 Pro
- U-DJI Neo
- Hover Air X1
Zonke lezi zixhobo ziqukethe imisebenzi yokulandelela okuzenzakalelayo esekelwe ekuboneni kwekhompyutha, ngokwethiyori, eyenzelwe ukwenza kube lula ukudubula okunamandla ngaphandle kokuba umshayeli wendiza akhathazeke ngekhamera. Yilokho kanye ikhono lokulandela umsebenzisi ngokuzimela Yilowo ophendukela kuye lapho kuvela iphethini efana neFlyTrap.
Ekuhlolweni, ama-drone adonswe ngokwanele ukwenza ukusebenza kwawo kube nokwenzeka. bamba ngamadivayisi enethiwekhi noma umthelela wayo olawulwayoNgamanye amazwi, kwakungeyona nje indaba yokudideka i-algorithm okwesikhashana, kodwa kwakuwukuphoqa indlela eqhubekayo engokomzimba eyashiya idivayisi esandleni somhlaseli.
Abacwaningi abaphethe ucwaningo benze ngokuvumelana nemikhuba emihle yomphakathi wezokuphepha kwe-inthanethi kanye Bakhuluma ngobuthakathaka ngendlela efanele. kubakhiqizi abathintekayo ngaphambi kokwenza yonke imininingwane ibe sobala. Lokhu kuvula ithuba lokuthi izinkampani ezifana ne-DJI nezinye izinkampani ezithintekile zibukeze ama-algorithms azo okulandelela futhi zisebenzise izindlela zokumelana nezibuyekezo ze-firmware zesikhathi esizayo noma izizukulwane zehadiwe.
Kodwa-ke, ngale kwalezi zinhlobo ezithile, ukutholwa kuye kwabonakala imiphumela ebanzi kakhuluNoma iyiphi i-UAS (uhlelo lwezindiza olungenamuntu) oluthembele kumanethiwekhi e-neural ukulandelela amatshe asetshenziswa ngokubona kwekhompyutha lungase lube sengozini yokushintshashintsha kwendlela efanayo uma lungaqeqeshwanga ngokuqondile ukumelana nalezi zinhlobo zamaphethini anonya.
Ukuhlaselwa ngokomzimba okungadingi umsakazo, ukugenca, noma ukuphazamiseka
Esinye sezici ezibalulekile zeFlyTrap ukuthi isebenza ngokuphelele isizinda esingokoqoboAkudingi ukuphazamisa ukuxhumana kwe-drone, ukufinyelela isixhumanisi sayo sedatha, noma ukusebenzisa ubuthakathaka besofthiwe yendabuko. Lokhu kuhlasela kusebenza ngoba kulawula lokho okubonwa yikhamera ye-drone, ngakho-ke, lokho ama-algorithms ayo e-AI akholelwa ukuthi kuyenzeka.
Empeleni, lokhu kusho ukuthi umhlaseli akudingeki akhiphe amasignali omsakazo, i-WiFi, i-GPS noma okufanayoFuthi akudingi noma yiluphi uhlobo lokuxhumana ne-drone, izimvume zokufinyelela, noma iziqinisekiso. Kuhilela nje ukukhombisa i-drone into ebonakalayo eyenzelwe ukusebenzisa izindawo zayo zokungaboni kahle, into enzima kakhulu ukuyilandela nokuyivimba ngezivikelo ezijwayelekile.
Le ndlela ifanelana nalokho okwaziwa ngokuthi ukuhlaselwa okuphikisayo ezweni elingokoqoboLolu uchungechunge locwaningo olubaluleke kakhulu. Njengoba nje kudalwa izibuko, ama-t-shirt, noma izitika ezingakhohlisa izinhlelo zokubona ubuso noma izimoto ezizimele, lapha isambulela sisetshenziswa ukulahla ama-algorithms okulandelela e-drone.
Ngaphezu kwalokho, izindleko ziphansi kakhulu uma kuqhathaniswa nezinsiza ezivame ukuhlotshaniswa nempi kagesi noma ukudilizwa kwezindiza ezingenamuntu. Akukho mishini ebizayo, ama-antenna akhethekile, noma ulwazi lwe-RF olungochwepheshe oludingekayo.Okudingayo nje isambulela esinomklamo ofanele kanye nekhono elithile lokuzibeka endaweni efanele lapho i-drone ikwimodi yokulandela.
Lokhu kungalingani phakathi kwentengo yedrone ethuthukisiwe kanye nezindleko zokuyinciphisa kusetshenziswa into evamile kunomthelela oqondile endleleni okufanele sicabange ngayo ukuphepha kwengqalasizinda ebalulekile, ukuthunyelwa kwamaphoyisa, noma imingceleNgokwenkolelo-mbono, umenzi wobubi angasondeza ama-drone okuqapha eduze kwezindawo ezivinjelwe esebenzisa lolu hlobo lwamasu ngqo, bese ewakhubaza noma awashaye lapho ethanda khona.
Imiphumela yokuphepha, ubumfihlo, kanye nokusetshenziswa kwama-drone amaningi
Ukwanda kwezindiza ezingenamuntu eziqhutshwa yi- ukuhlakanipha okufakelwayo Lokhu kubeka inselele enkulu ekuphepheni komphakathi. Sibona amaphrojekthi amaningi aphakamisa ukusebenzisa izinqwaba zama-drone ukuqapha amadolobha, ukuqapha imingcele, noma ukuqondisa imicimbi emikhulu. Kodwa ucwaningo lwe-UC Irvine lubonisa ukuthi, ngaphandle kokuqinisa ama-algorithms okuqonda, konke lokhu kusetshenziswa kusekelwe esisekelweni esibuthakathaka.
Ezingeni lokusebenza, ukuhlasela kwe-FlyTrap kungasetshenziswa kokubili ekuhlaseleni nasekuzivikeleni. Ngakolunye uhlangothi, othile angase vimbela amaphoyisa noma ama-drone aqapha umngceleLokhu kubeka engcupheni amandla okuphendula amabutho ezokuphepha endaweni ethile. Ngakolunye uhlangothi, umuntu ohlukunyezwa yi-drone, noma isisulu sobunhloli ngedivayisi yezentengiselwano, angasebenzisa indlela efanayo ukuzivikela ngesambulela esilula esilungisiwe.
Kukhona futhi isici se- ukuvikelwa kwengqalasizinda yamasu (iziteshi zikagesi, izikhungo zezokuthutha, izikhungo zedatha, njll.). Uma lezi zindawo zithembele kuma-drones azimele ukuze zengeze ezinhlelweni zazo zokuqapha, ubuthakathaka obufana ne-FlyTrap buvula ingozi yokuthi umhlaseli angadlula ingxenye yesistimu yokuthola esebenzisa into ebonakalayo kuphela ngaphandle kokuxhumana kwe-elekthronikhi.
Ngakho-ke lolu cwaningo luthumela umyalezo ocacile: Ukuphepha kwe-drone akukwazi ukukhawulelwa kungqimba ye-elekthronikhi noma yenethiwekhiUkuvikela izixhumanisi zedatha, ukubethela ukuxhumana, nokuvikela isofthiwe kuyadingeka, kodwa akwanele, uma ama-algorithms okubona ekhohliswa amaphethini anemibala aphrintiwe esambuleleni. Ukuqina kumele kudlulele endleleni i-AI ehumusha ngayo indawo ebonakalayo kanye nokuqina kwayo ekuphathweni ngamabomu.
Njengoba ukusetshenziswa kwama-drone azimele kusabalala kakhulu izindawo zasemadolobheni kanye nemisebenzi ebucayiLezi zinhlobo zokuhlaselwa ngeke zisaba yinto ethakazelisayo yezemfundo futhi zizoba yisici okuzodingeka abakhiqizi, abalawuli, kanye nabaqhubi basicabangele kusukela esigabeni sokuklama. Ukungazinaki lezi zibuthakathaka kungaba yindleko enkulu uma ukuthunyelwa okukhulu sekuvele kwenzeke.
Ukulinganiselwa kokuhlasela kanye nemigqa yokuzivikela engaba khona
Nakuba umqondo "wokudubula i-drone ngesambulela" ukhanga kakhulu, akufanele ucabange ukuthi noma yisiphi isambulela esinemibala sizokwanelisa. Iphethini ye-FlyTrap iklanywe ngokucophelela ukusebenzisa ubuthakathaka obuthile kuma-algorithms ahlolwe. Akuyona icebo elijwayelekile elizosebenza ngokuzenzakalelayo nawo wonke ama-drone noma kuzo zonke izimo.
Ngaphezu kwalokho, ukuhlasela kudinga ukuthi i-drone isebenzise imisebenzi ye ukulandelela okusekelwe embonweni ozimeleUma umqhubi eqhuba ngesandla, noma uma indiza iqondiswa ezinye izinzwa (njenge-LIDAR, i-radar, noma inhlanganisela ye-sensor fusion ethuthukisiwe), umkhawulo wokukhohlisa i-AI usebenzisa iphethini ebonakalayo kuphela ungancishiswa noma unyamalale.
Omunye umkhawulo osebenzayo yisidingo sokuthi sondeza isambulela eduze nensimu yokubona yedrone ngokomzimbaEzimweni lapho idivayisi indiza endaweni ephakeme noma igcina ibanga elingamakhilomitha amaningana ukusuka endaweni okuqondiswe kuyo, ukusondela lapho kungase kungabi yinto engokoqobo. I-FlyTrap iyingozi kakhulu ezimweni lapho indiza iphansi, ukulandelela abantu eduze, noma izindawo zasemadolobheni ezinemizila emincane.
Kusukela ohlangothini lokuzivikela, kunezindlela eziningana zokwenza. Omunye wazo uwukuthi qeqesha kabusha amamodeli okubona Besebenzisa izibonelo zamaphethini aphikisanayo njengeFlyTrap, ukuze bafunde ukuwabona njengeziphambeko futhi bagweme ukuwela ogibeni lokusondela ngokungalawuleki. Enye inketho ukuhlanganisa ulwazi olubonakalayo neminye imithombo (ukujula, izinzwa ezingasebenzi, amamephu e-3D) ukuze bagweme ukuthembela kuphela kulokho ikhamera ekubonayo.
Kunengqondo futhi ukuyifaka kwi-firmware imikhawulo yokusondela kanye nemithetho yokuphepha eyengeziweLezi zinyathelo zivimbela i-drone ukuthi isondele eduze kwesithako esiseduze nebanga elithile uma ibona izinguquko ezibonakalayo ezingavamile noma ezingaguquki. Lokhu kungase kungayiqedi ngokuphelele inkohliso, kodwa kunganciphisa amathuba okuba i-drone ifinyelele endaweni lapho ingabanjwa khona kalula ngesandla noma ngenethi.
Ekugcineni, ezingeni lokulawula, izinhlangano ezinesibopho sokuqinisekisa ama-drone okusetshenziswa kwawo ukuqapha ezokuphepha, amaphoyisa noma ingqalasizinda Kuzodingeka bafake izivivinyo ezimelene nalolu hlobo lokuhlaselwa ngokomzimba ekuhlolweni kwabo. Ukusebenzisa i-drone ukuqopha indawo yezintaba akufani nokuyisebenzisa ukugada umngcele: izinga lokuqina elidingekayo kufanele lihluke ngokusobala.
Konke lokhu kushiya isimo esimangalisayo: lapho senza ama-drone “ngokuhlakanipha” okwengeziwe nge-AI, Okubaluleke kakhulu kuzoba ukucabanga njengomhlaseli odala ozama ukuthola izindlela ezinqamulelayo ezweni langempela ukuze abakhohlise. Futhi yilapho into ebonakala ingenacala njengesambulela esinemibala ingagcina isiyi-protagonist.
Ngamafuphi, okuboniswa yilezi zifundo ukuthi Ukuphepha kwezinhlelo ezizimele akuyona nje indaba yokuvikela umlilo kanye nokubethelakodwa futhi nokuqonda ukuthi babona kanjani imvelo, ukuthi yiziphi izindlela ezinqamulelayo ezisetshenziswa yi-algorithms yabo, nokuthi iphethini elula yesithombe ingaguqula kanjani indiza ethuthukisiwe engenamuntu ibe yinto esengozini njengempukane ekhangwa ugibe oluklanywe kahle.