Zaɓi Harshe

Wani Sabon Tsarin Gano Matsayin Motoci bisa Haɗakar Sadarwa ta Kamara ta Haske da Hotunan Zane

Wani sabon hanyar gano matsayin mota ta amfani da sadarwar fitilun baya da hotunan zane don motocin da ke gudanar da kansu, yana inganta daidaito ba tare da manyan sauye-sauyen ababen more rayuwa ba.
rgbcw.org | PDF Size: 0.8 MB
Kima: 4.5/5
Kimarku
Kun riga kun ƙididdige wannan takarda
Murfin Takardar PDF - Wani Sabon Tsarin Gano Matsayin Motoci bisa Haɗakar Sadarwa ta Kamara ta Haske da Hotunan Zane

1. Gabatarwa

Daidaitaccen gano matsayin mota shine ginshiƙi don amintaccen amfani da motocin da ke gudanar da kansu (AVs). Duk da yake Tsarin Tauraron Dan Adam na Duniya (GNSS) kamar GPS sun zama ruwan dare gama gari, suna fama da raguwar siginar a cikin kwaruruka na birane, ramuka, da ƙarƙashin ciyayi masu yawa, wanda ke sa su zama marasa dogaro ga ayyukan AV masu mahimmanci na aminci. Wannan takarda ta magance wannan gibi ta hanyar gabatar da wani sabon tsari mai sauƙi na gano matsayi wanda ke haɗa Sadarwar Kamara ta Haske (OCC) da hotunan zane tare.

Babban dalili ya samo asali ne daga ƙididdiga masu ban tsoro na mace-macen zirga-zirgar hanya, wanda galibi ana danganta su da haɗuwa cikin sauri. Fasahar tuƙi ta kai da kai tana alƙawarin rage wannan, amma tasirinta yana da alaƙa kai tsaye da sanin daidaitaccen matsayi. Hanyar da aka gabatar tana nufin samar da ƙarin mataki na gano matsayi ko madadin matsayi wanda yake mai sauƙi, amintacce, kuma yana amfani da kayan aikin mota da ake da su (fitilun baya, kamarori) tare da ƙaramin gyara ga ababen more rayuwa na waje.

1.1 Magungunan da ake da su, Iyakoki, da Yanayin Halin Yanzu

Gano matsayin mota na yanzu ya dogara da farko akan haɗakar firikwensin: haɗa GPS tare da Rukunin Ma'aunin Inertial (IMUs), LiDAR, radar, da hangen nesa na kwamfuta. Duk da yake yana da tasiri, wannan hanyar sau da yawa tana da rikitarwa da tsada. Hanyoyin da suka dogara kawai da hangen nesa na iya zama mai ƙarfin lissafi kuma sun dogara da yanayi. Hanyoyin da suka dogara da sadarwa kamar Sadarwar Gajeren Zango na Keɓaɓɓe (DSRC) ko Cellular-V2X (C-V2X) suna buƙatar keɓaɓɓen kayan aikin rediyo kuma suna da saukin kamuwa da tsangwama na RF da barazanar tsaro kamar yaudara.

Yanayin yana motsawa zuwa tsarin da yawa, tsarin da ya wuce kima. Sabon abu a nan shine amfani da fitilar baya na mota a matsayin mai watsa bayanai da aka daidaita (OCC) da kamarar motar da ke biye a matsayin mai karɓa, ƙirƙirar hanyar sadarwa kai tsaye ta V2V mai hangen nesa. Ana ƙara wannan ta hanyar amfani da fitilun titi (SLs) a matsayin wuraren tunani da aka sani ta hanyar hotunan zane, ƙirƙirar tsarin tunani na gauraye mai motsi da tsayayye.

Babban Dalili: Amincin Hanya

~1.3 miliyan mutuwar zirga-zirga a duk duniya a kowace shekara (WHO). Haɗuwa cikin sauri (>80 km/h) yana lissafin ~60% na mace-mace. Daidaitaccen gano matsayi yana da mahimmanci don guje wa haɗuwa a cikin AVs.

2. Tsarin Gano Matsayi da Ake Shawarar

2.1 Tsarin Tsarin da Rarraba Motoci

Tsarin ya gabatar da rarrabuwa mai sauƙi amma mai tasiri:

  • Mota Mai Masaukin (HV): Motar da ke aiwatar da gano matsayi. Tana da kamara kuma tana sarrafa siginar don kimanta matsayin wasu.
  • Mota Mai Ci Gaba (FV): Mota da ke motsawa a gaban HV. Tana watsa siginar tantancewa / yanayi ta hanyar fitilun baya ta amfani da OCC.
  • Fitilar Titi (SL): Ababen more rayuwa na tsaye tare da sanannun ma'auni, ana amfani da su azaman anka na matsayi cikakke don daidaita matsayin HV da kansa da rage kuskuren tarawa.

Kamarar HV tana aiki da manufa biyu: 1) a matsayin mai karɓar OCC don warware bayanai daga fitilar baya na FV, da 2) a matsayin firikwensin hotunan zane don auna nisa.

2.2 Tsarin Algorithm na Asali na Gano Matsayi

Algorithm yana aiki a cikin tsarin dangi kafin ya kafa ma'auni zuwa ma'auni cikakke:

  1. Gano Matsayin HV da kansa: HV yana amfani da hotunan zane don auna nisansa zuwa SLs biyu ko fi da aka sani. Ta hanyar kwatanta canjin waɗannan nisa yayin da yake motsawa, zai iya yin triangulate da inganta matsayinsa na cikakke akan taswira.
  2. Gano Matsayin Dangi na FV: A lokaci guda, HV yana amfani da hotunan zane don auna nisan dangi zuwa FV da ke gaba ta hanyar nazarin girman (pixels da aka mamaye) na fitilar baya na FV ko bayanan bayansa akan firikwensin hotonsa.
  3. Haɗakar Bayanai & Matsayin Cikakke: Siginar OCC da aka daidaita daga FV ya ƙunshi mai tantancewa na musamman. Da zarar HV ya san matsayinsa na cikakke (daga SLs) da daidaitaccen vector na dangi zuwa FV (daga hotunan zane), zai iya lissafin matsayin cikakke na FV.

Babban sabon abu shine kwatanta saurin canji na nisa tsakanin HV-SL da HV-FV. Wannan bincike na bambance-bambance yana taimakawa tace kurakurai na gama gari kuma yana inganta ƙarfi.

Babban Fahimta

  • Firikwensin Amfani Biyu: Ana amfani da kamara don duka sadarwa (OCC) da fahimta (hotunan zane), yana haɓaka amfanin kayan aiki.
  • Ababen More Rayuwa Mai Sauƙi: Ya dogara da fitilun titi da fitilun mota da ake da su, yana guje wa ƙaddamar da sabbin ababen more rayuwa masu yawa.
  • Aminci na Asali: Yanayin hangen nesa na OCC yana sa ya yi wahala a yaudara ko toshe shi daga nesa idan aka kwatanta da siginar RF.

3. Cikakkun Bayanai na Fasaha & Tushen Lissafi

Lissafin nisa na hotunan zane shine tsakiyar tsarin. Babban ƙa'idar ita ce girman abu da aka sani a cikin jirgin hoto yana daidaitawa da nisansa daga kamara.

Dabarar Kimanta Nisa: Don abu mai sanannen tsayi na ainihi $H_{real}$ da faɗi $W_{real}$, nisa $D$ daga kamara ana iya kimanta shi ta amfani da ƙirar kamara ta rami: $$D = \frac{f \cdot H_{real}}{h_{image}} \quad \text{ko} \quad D = \frac{f \cdot W_{real}}{w_{image}}$$ inda $f$ shine tsayin mai da hankali na kamara, kuma $h_{image}$ da $w_{image}$ sune tsayi da faɗin abu a cikin firikwensin hoto (a cikin pixels), wanda aka daidaita zuwa raka'a na zahiri.

Daidaitawar OCC: Fitilar baya na FV (mai yiwuwa tsararrun LED) ana daidaita ta a mitar da ta isa ba za a iya gani da idon ɗan adam ba amma ana iya gano ta ta hanyar kamara mai rufewa ko mai rufewa ta duniya. Dabarori kamar Maɓalli Kan-Kashe (OOK) ko Maɓalli Canjin Launi (CSK) ana iya amfani da su don ɓoye ID na mota da bayanan motsi na asali.

Dabarar Haɗakar Bayanai: Bari $\Delta d_{SL}$ ya zama canjin nisa da aka auna tsakanin HV da Fitilar Titi na tunani, kuma $\Delta d_{FV}$ ya zama canjin nisa da aka auna tsakanin HV da FV. Idan matsayin HV da kansa an san shi daidai, waɗannan canje-canjen ya kamata su yi daidai da ƙuntatawa na geometric. Ana amfani da bambance-bambance don gyara kimantawar matsayin dangi na FV da kimantawar yanayin HV da kansa a cikin tsarin tacewa (misali, Tace Kalman).

4. Sakamakon Gwaji & Binciken Aiki

Takardar ta tabbatar da tsarin da aka gabatar ta hanyar gwajin auna nisa, muhimmin mataki na farko.

Bayanin Taswira & Sakamako: Duk da yake abin da aka fitar na PDF bai nuna takamaiman jadawali ba, rubutun ya bayyana cewa sakamakon gwaji "ya nuna gagarumin ci gaba a cikin aiki" kuma "gwajin auna nisa ya tabbatar da yuwuwar." Muna iya fahimtar ma'auni na yiwuwar aiki da nau'ikan jadawali:

  • Kuskuren Kimanta Nisa vs. Nisa na Gaskiya: Taswira mai layi wanda ke nuna kuskuren cikakke a mitoci na kimanta nisa na hotunan zane ga duka SLs da FVs a cikin kewayon (misali, 5m zuwa 50m). Ana sa ran kuskuren zai ƙaru tare da nisa amma ya kasance cikin iyaka, kewayon da aka yarda don aikace-aikacen mota (mai yiwuwa ƙasa da mita a kewayon da ya dace).
  • Daidaitaccen Gano Matsayi CDF (Aikin Rarraba Tarawa): Jadawali da ke zana yuwuwar (y-axis) cewa kuskuren gano matsayi ya kasance ƙasa da wani ƙima (x-axis). Lanƙwasa mai tsayi yana jujjuyawa zuwa hagu yana nuna babban daidaito da daidaito. Hanyar gauraye da aka gabatar (OCC+Hotunan Zane+SL) zai nuna lanƙwasa wanda ya fi amfani da hotunan zane kawai ko OCC na asali ba tare da anka na SL ba.
  • Aiki a Ƙarƙashin Yanayi daban-daban: Jadawali na sanduna yana kwatanta ma'auni na kuskure a cikin yanayi daban-daban: rana/dare, yanayi mai tsabta/ruwan sama, tare/ba tare da bayanan tunani na SL ba. Za a nuna ƙarfin tsarin ta hanyar kiyaye aiki mai ɗan kwanciyar hankali, musamman lokacin da bayanan SL suke samuwa.

Babban abin da za a ɗauka shine cewa hanyar haɗakar tana rage raunin kowane ɓangare: OCC yana ba da ID, hotunan zane yana ba da kewayon dangi, kuma SLs suna ba da wuraren anka cikakke.

5. Tsarin Bincike: Nazarin Lamari Ba tare da Lambar Ba

Yanayi: Babbar hanya mai layi uku da dare. HV yana cikin layin tsakiya. FV1 yana gaba kai tsaye a cikin layi ɗaya. FV2 yana cikin layin hagu, ɗan kaɗan a gaba. Fitilun titi guda biyu (SL1, SL2) suna kan gefen hanya tare da sanannun ma'auni na taswira.

Tsarin Gano Matsayi Mataki-mataki:

  1. Fara Aiki: Tsarin HV yana da taswira mai ɗauke da matsayin SL1 da SL2.
  2. Gano Matsayin HV da kansa: Kamarar HV ta gano SL1 da SL2. Ta amfani da hotunan zane (sanin daidaitattun girmomin fitilar titi), tana lissafin nisa $D_{HV-SL1}$ da $D_{HV-SL2}$. Ta hanyar daidaita waɗannan nisa da kusurwoyi zuwa taswira, tana ƙididdige daidaitattun ma'auni $(x_{HV}, y_{HV})$ da kanta.
  3. Gano FV & Sadarwa: Kamarar HV ta gano hanyoyin fitila biyu na baya (FV1, FV2). Tana warware siginar OCC daga kowanne, yana samun IDs na musamman (misali, "Veh_ABC123", "Veh_XYZ789").
  4. Auna Nisa na Dangi: Ga kowane FV, ana amfani da hotunan zane akan tarin fitilun baya (sanannen girman tsararrun LED) don ƙididdige nisan dangi $D_{rel-FV1}$ da $D_{rel-FV2}$, da kusurwar ɗauka.
  5. Matsayi Cikakke: HV yanzu yana haɗa matsayinsa na cikakke $(x_{HV}, y_{HV})$ tare da vector na dangi $(D_{rel}, \theta)$ ga kowane FV. $$(x_{FV}, y_{FV}) = (x_{HV} + D_{rel} \cdot \sin\theta, \, y_{HV} + D_{rel} \cdot \cos\theta)$$ Wannan yana haifar da matsayin taswira cikakke na FV1 da FV2.
  6. Tabbatarwa & Bi: Yayin da duk motoci ke motsawa, ana sa ido kan ci gaba da canji a cikin $\Delta d_{SL}$ da $\Delta d_{FV}$. Rashin daidaituwa yana haifar da daidaita maki amincewa ko sabunta tacewa, yana tabbatar da bi mai santsi da dogaro.
Wannan tsarin yana nuna yadda tsarin ke ƙirƙirar taswira na yanki, mai motsi na zirga-zirgar da ke kewaye ta amfani da ƙaramin musayar bayanai.

6. Bincike Mai Zurfi & Ra'ayi na Kwararru

Babban Fahimta: Wannan takarda ba kawai wata takarda ta haɗakar firikwensin ba ce; yana da wayo sake amfani da kayan aiki. Marubutan sun gano cewa fitilar baya ta LED da kamara—ɓangarori biyu na gama gari, waɗanda aka tilasta su akan motoci na zamani—za a iya canza su zuwa tsarin sadarwa da auna nisa na V2V mai aminci, ƙarancin bandwidth tare da sabunta software. Wannan yana rage matsalar shiga sosai idan aka kwatanta da ƙaddamar da sabbin rediyon V2X na tushen RF.

Kwararar Hankali & Kyakkyawan Fahimta: Hankali yana da kyau da madauwari kuma yana gyara kansa. HV yana amfani da alamomin tsaye (SLs) don nemo kansa, sannan yana amfani da kansa don nemo abubuwa masu motsi (FVs). Hanyar haɗin OCC yana ba da tantancewa tabbatacce, yana magance matsalar "haɗakar bayanai" wanda ke addabar hangen nesa na kwamfuta kawai (misali, "wannan mota ɗaya ce da na gani firam biyu da suka wuce?"). Amfani da hotunan zane akan sanannen tushen haske, wanda aka sarrafa (fitilar baya) ya fi dogaro da ƙoƙarin kimanta nisa zuwa siffar mota na gaba ɗaya, wanda zai iya bambanta sosai. Wannan yana tunawa da yadda Alamun Afrilu ko Alamomin ArUco ke aiki a cikin injinan mutum-mutumi—amfani da tsari da aka sani don daidaitaccen kimanta matsayi—amma ana amfani da shi da sauri a cikin mahallin mota.

Ƙarfi & Kurakurai:

  • Ƙarfi: Mai Tsada & Mai Wadatarwa: Babban nasara. Babu sabon kayan aiki don motoci ko hanyoyi a cikin mafi kyawun yanayi. Aminci: Hangen nesa na zahiri shine ƙa'ida mai ƙarfi ta tsaro. Kiyaye Sirri: Ana iya ƙirƙira shi don musayar ƙaramin bayani, bayanan da ba su tantance ba. Banda Bakan RF: Ba ya gasa don cunkoson bandakunan rediyo.
  • Kurakurai & Tambayoyi: Hankali na Muhalli: Ta yaya yake aiki a cikin ruwan sama mai ƙarfi, hazo, ko dusar ƙanƙara wanda ke watsa haske? Shin kamara za ta iya gano siginar da aka daidaita a ƙarƙashin hasken rana mai haske ko kuma glare? Iyakacin Kewayon: OCC da hotunan zane na tushen kamara suna da iyakacin kewayon tasiri (mai yiwuwa <100m) idan aka kwatanta da radar ko LiDAR. Wannan yana karɓuwa don gano barazana nan take amma ba don tsarin shirye-shirye na dogon zango ba. Dogaro akan Ababen More Rayuwa: Duk da yake "ababen more rayuwa mai sauƙi," har yanzu yana buƙatar SLs tare da sanannun ma'auni don mafi kyawun daidaito. A yankunan karkara ba tare da irin waɗannan SLs ba, daidaito yana raguwa. Nauyin Lissafi: Sarrafa hoto na ainihin lokaci don hanyoyin haske da yawa da hotunan zane ba abu ne mai sauƙi ba, kodayake ci gaban na'urori na musamman na hangen nesa (kamar na NVIDIA ko Mobileye) suna rufe wannan gibi.

Fahimta Mai Aiki:

  1. Ga Masu Kera Motoci: Wannan ya kamata ya kasance kan taswirar hanya a matsayin ƙarin mataki na aminci mai ƙari. Fara ƙirƙira ta hanyar daidaita zagayowar aiki na LED a cikin fitilun baya da amfani da kamarorin kewaye da ake da su. Daidaita ƙa'idar OCC mai sauƙi don IDs na mota shine 'ya'yan itace masu sauƙi ga ƙungiyoyi kamar AUTOSAR ko IEEE.
  2. Ga Masu Tsara Birane: Lokacin shigarwa ko haɓaka fitilun titi, haɗa alama mai sauƙi, mai karantawa ta inji (kamar tsarin QR) ko tabbatar da an daidaita girmansu kuma an yi rajista a cikin taswirori masu ƙima. Wannan yana mai da kowane sandar fitila ya zama fitilar gano matsayi kyauta.
  3. Ga Masu Bincike: Mataki na gaba shine haɗa wannan yanayin cikin cikakken kayan aikin firikwensin. Ta yaya yake haɗawa da radar 77GHz a cikin rashin gani? Shin za a iya haɗa bayanansa tare da guguwar batu na LiDAR don inganta rarraba abu? Bincike ya kamata ya mayar da hankali kan algorithms masu ƙarfi don mummunan yanayi da kuma kwatanta da V2X na tushen RF a cikin yanayin guje wa haɗuwa na ainihi, kama da binciken da Ma'aikatar Sufuri ta Amurka ta gudanar don DSRC.
Wannan aikin mataki ne mai ma'ana zuwa ga dimokuradiyya daidaitaccen gano matsayi. Ba zai maye gurbin babban LiDAR ba amma zai iya sanya gano matsayi "mai kyau" don ayyuka da yawa na AV ya zama samuwa ga motoci da yawa, da sauri.

7. Aikace-aikace na Gaba & Hanyoyin Bincike

1. Ƙungiya da Gudanar da Cruise na Haɗin Kai (CACC): Daidaitaccen, ƙananan matsayi na dangi wanda wannan tsarin ya ba da dama yana da kyau don kiyaye ƙungiyoyin mota masu ƙarfi, masu ingancin mai a kan manyan hanyoyi. Hanyar haɗin OCC na iya watsa ƙaddamarwa/rage sauri kai tsaye daga fitilun birki na motar jagora.

2. Ƙarfafawa don Kariyar Masu Amfani da Hanya masu Rauni (VRU): Keke, scooter, da masu tafiya a ƙasa za a iya sanya su da ƙananan alamun LED masu aiki waɗanda ke watsa matsayinsu da yanayin tafiyarsu ta hanyar OCC. Kamarar mota za ta gano waɗannan alamun ko da a cikin hangen nesa na gefe ko da dare, yana samar da ƙarin matakin aminci fiye da na al'ada na firikwensin.

3. Gano Matsayi a Ciki & Ƙarƙashin Ƙasa: A cikin wuraren da aka hana GPS kamar gidajen ajiye motoci masu hawa da yawa, ramuka, ko tashoshi, fitilun LED da aka daidaita a cikin rufin na iya zama masu watsa OCC suna watsa ma'auni na cikakke. Motoci na iya amfani da wannan don daidaitaccen gano matsayi don nemo wuraren ajiye motoci ko kuma su yi tafiya da kansu a cikin filayen dabaru.

4. Haɗawa da Taswirori HD da SLAM: Tsarin na iya samar da sabuntawa na ainihin lokaci, matsayi cikakke don gyara karkata a cikin Tsarin Gano Matsayi da Taswira (SLAM) da AVs ke amfani da su. Kowane mota da aka gano matsayi ya zama ma'anar bayanai wanda zai iya tara jama'a don sabunta taswira HD (misali, bayar da rahoton yankin gini na wucin gadi).

5. Daidaitawa da Tsaron Sadarwa: Aikin nan gaba dole ne ya mayar da hankali kan daidaita tsare-tsaren daidaitawa, tsarin bayanai, da ka'idojin tsaro (misali, sirri mai sauƙi don tabbatar da saƙo) don hana hare-haren yaudara inda mai aikata mugunta yake amfani da LED mai ƙarfi don kwaikwayi siginar mota.

8. Nassoshi

  1. Hossan, M. T., Chowdhury, M. Z., Hasan, M. K., Shahjalal, M., Nguyen, T., Le, N. T., & Jang, Y. M. (Shekara). A New Vehicle Localization Scheme based on Combined Optical Camera Communication and Photogrammetry. Sunan Jarida/Taron.
  2. Hukumar Lafiya ta Duniya (WHO). (2023). Rahoton Matsayin Duniya kan Amincin Hanya. Geneva: WHO.
  3. Ma'aikatar Sufuri ta Amurka. (2020). Shirin Ƙaddamar da Jirgin Ruwa mai Haɗin kai: Rahoton Kimantawa na Mataki na 2. An samo daga [Gidan Yanar Gizon USDOT].
  4. Zhu, J., Park, J., & Lee, H. (2021). Robust Vehicle Localization in Urban Environments Using LiDAR and Camera Fusion: A Review. IEEE Transactions on Intelligent Transportation Systems.
  5. Caesar, H., et al. (2020). nuScenes: A multimodal dataset for autonomous driving. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).
  6. IEEE Standard for Local and metropolitan area networks–Part 15.7: Short-Range Wireless Optical Communication Using Visible Light. (2018). IEEE Std 802.15.7-2018.