Imephu ehlakaniphile ye Ukuzulazula okufinyelelekayo
Imephu ye-sensor-fusion yasendlini ethuthukisiwe eshintsha izikhala eziyinkimbinkimbi zibe izindawo okuzula kuzo abantu abayizimpumputhe noma abangaboni kahle.
Ubuchwepheshe bokuhlanganisa izinzwa
Sihlanganisa ubuchwepheshe obuningi ukuze sakhe isistimu yokuzulazula yasendlini enembe kakhulu, efinyelelekayo emakethe
I-geolocation kaMazibuthe (I-MagneticField)
Inkambu kazibuthe eyingqayizivele yesakhiwo ngasinye isebenza "njengezigxivizo zeminwe zedijithali." Izakhiwo zikakhonkolo nezensimbi zakha iphethini ehlukile esebenza njengekhodi yebha engabonakali yokuma okunembile ngaphandle kwe-GPS.
I-Wi-Fi RTT ne-Bluetooth LE
I-Wi-Fi Round-Trip Time (RTT) ikala isikhathi sokuya nokubuya ukuze sibe isentimitha ukunemba. Amabhikhoni e-Bluetooth Low Energy (BLE) engeza amaphoyinti okubhekisela kunxantathu wesikhathi sangempela.
I-Inertial sensor fusion (IMU)
Isilinganiso esingenasici sihlanganisa i-accelerometer (ukunyakaza), i-gyroscope (ukuzungezisa), i-magnetometer (isihloko), kanye ne-barometer (ukuphakama/phansi) ukuze kube nokubala okuthembekile okufile.
Umbono wekhompyutha & AR
Ukubona kwekhompyutha kuthola izici ezibalulekile (i-SLAM), izimpawu, amakhodi e-QR, nezimaka ze-AR ukuze kulungiswe indawo futhi kusekelwe ukuzulazula okuthuthukisiwe ngesikhathi sangempela.
Ukunemba kwemitha yokugcina (99%)
Ukuhlanganiswa kwethu kwezinzwa ezinelungelo lobunikazi kuqinisekisa 99% ukunemba phakathi kwamamitha angu-1–2, okubalulekile ekuzulazuleni okuphephile kubantu abayizimpumputhe. Ngenkathi i-GPS isebenza ngaphandle, isistimu yethu ihamba phambili ngaphakathi endlini lapho ukuzimela kuyinselelo enkulu khona. Ukubambezeleka kwe-Sub-100ms kusho impendulo yesikhathi sangempela.
Izici zokufinyeleleka
Isici ngasinye sakhelwe izidingo zangempela zabantu abayizimpumputhe noma abangaboni kahle
I-Multimodal sensor fusion (izendlalelo ezi-6)
Isitaki sokuhlanganisa esinelungelo lobunikazi esihlanganisa ukuma kazibuthe, i-Wi-Fi RTT, i-Bluetooth LE, i-IMU (i-accelerometer + gyroscope + barometer), ukubona kwekhompyutha, ne-GNSS. Isihlungi sezinhlayiyana sihlanganisa zonke izinzwa ngokuhambisana ukuze kube nokunemba okungu-99%.
Ukuma kwesikhathi sangempela (sub-100ms)
Ukubambezeleka okungaphansi kuka-100ms ngezibuyekezo zesimo eziqhubekayo. Unqenqema olukudivayisi kanye nefu lokusebenza okuphezulu. Ukubala okufile kubamba ukunemba lapho amasignali angaphandle ehla.
Ingqalasizinda encane (Izigxivizo Zeminwe zikaMagnetic)
Izigxivizo zeminwe kazibuthe ezinelungelo lobunikazi zidinga ibhekhoni kumakesi angu-80%. Sonke isakhiwo sikakhonkolo nesensimbi sikhiqiza iphethini kazibuthe eyingqayizivele esebenza njengekhodi yebha engabonakali yasendlini.
Ukufunda komshini oguquguqukayo
Amanethiwekhi e-Neural aqeqeshwe ku-TensorFlow Lite afunda amaphethini okunyakaza, izintandokazi zabasebenzisi, nezici zendawo. Uhlelo luhlala luthuthuka kudivayisi nokufunda ngokuhlanganyela.
Umsindo wendawo we-3D (i-synthetic echolocation)
Umsindo we-Binaural ushintsha ibanga nesiqondiso kumsindo we-spatialized. I-HRTF idala “i-sonar” yokwenziwa ukuze abasebenzisi abayizimpumputhe bakwazi “ukubona” ngezinkambu zomsindo ezingu-360°.
Umbono wekhompyutha + SLAM
I-SLAM ithola izici, izimpawu, amakhodi e-QR, nezithiyo ngesikhathi sangempela. Izinhlobo ze-YOLO ne-MobileNet zisebenza kudivayisi ngo-30 FPS.
Izinzuzo kubasebenzisi
Ukuguqula izinselele zibe ukuzimela nokuzimela
Isiqondiso esizenzakalelayo sokujika nejika
Isiqondiso sesinyathelo nesinyathelo esinomsindo we-3D we-spatialized, ukuze abantu abayizimpumputhe bakwazi ukuzulazula ezindaweni eziyinkimbinkimbi—izitolo ezinkulu, izikhumulo zezindiza, izibhedlela, namakhempasi—ngokuzethemba.
Ukutholwa kwesithiyo ngesikhathi sangempela
Ihlanganisa i-LiDAR, ikhamera ejulile, nezinzwa ze-ultrasonic ukuze ihlabe umkhosi izithiyo—izinto zesikhashana, abantu, izinyathelo, noshintsho olungaphezulu—endleleni.
Amaphoyinti ahlakaniphile athakaselayo
Isizindalwazi se-georeferenced esinezinkulungwane zama-POI: izindlu zangasese ezifinyelelekayo, amakheshi, izitebhisi ezizihambelayo, izindawo zokuphuma eziphuthumayo, izinto zokubala zesevisi, ama-ATM, nokuningi.
Ukuthola indlela okufinyelelekayo
I-A* elungisiwe elungiselela ukufinyeleleka: igwema izitebhisi, ikhetha amakheshi, icabangela ububanzi bephaseji, izindlela ezithintekayo, nesikhathi sokuwela.
Isebenza kanjani imephu
Ukuphela ukuya ekugcineni, ukusuka ekuthwebuleni kuya ekuhambeni bukhoma kubasebenzisi
Ukuthwebula idatha
Sihamba kusayithi siqoqa idatha kazibuthe, i-Wi-Fi, ne-Bluetooth ukuze sakhe imephu yedijithali enothile.
Ukucubungula kwe-AI
Amamodeli e-AI akhipha amaphethini ahlukile futhi akhe imodeli enembile yemvelo.
Ukuqinisekisa okufinyelelekayo
Abasebenzisi abayizimpumputhe nabangaboni kahle bahlola imephu ukuze imizila ifinyeleleke ngempela futhi iphephe.
Izibuyekezo eziqhubekayo
Uhlelo lugcina ukufunda, luthuthukisa ukunemba nokujwayela izinguquko zesakhiwo.
Izindawo ezimephuwe
Siletha ukufinyeleleka ezindaweni eziyinselele kakhulu
Izikhungo zokuthenga
Hamba phakathi kwezitolo, thola izindlu zangasese ezifinyeleleka kalula, thola amakheshi, futhi uthole okunikezwayo okuklanyelwe.
Izikhumulo zezindiza
Thola isango lakho, thola amasevisi, futhi ufike esangweni ngesikhathi ngokuzethemba okugcwele.
Izibhedlela
Thola imitholampilo, amalebhu, amakhemisi, namadeski angaphambili ezindaweni zokunakekela eziyinkimbinkimbi.
Amanyuvesi
Thola amakilasi, imitapo yolwazi, amahholo okudlela, futhi uzungeze ikhempasi ngokuzimela.