ܭଌࣗಈ੍ ֶޚձจू Vol.xx, No.xx, 1/7ʢ20xxʣ 3 ࣍ݩϚʔΧΛ༻͍ͨϏδϡΞϧαʔϘܕਫதϩϘοτͷؾ๐֎ཚʹର͢Δ ੍ޚಛੑ ೲ ཅ∗ ɾถ ݟ ࿘ ݈ ଠ∗ ɾੴ ࢁ ৽ଠ∗ ∗ޢɾদ ོ ∗ Control characteristics of visual-servo type underwater vehicle system using three-dimensional marker for air bubble disturbance Akira YANOU∗ , Kenta YONEMORI∗ , Shintaro ISHIYAMA∗ , Mamoru MINAMI∗ and Takayuki MATSUNO∗ This paper studies a control characteristics of visual-servo type underwater vehicle system using threedimensional (3D) marker under air bubble disturbance on real-time pose tracking for visual servoing. The recognition of vehicle’s pose through three-dimensional marker is executed by Genetic Algorithms (GA). The proposed system does not merely calculate the position and orientation information, but can recognize the target position and orientation information through GA while visual servoing, because the system utilizes a three-dimensional marker shape and color to recognize the marker. In our previous research, a regulator performance of the system about the vehicle’s pose to 3D marker is explored under the condition that there is no disturbance on images. Therefore this paper explores the control results the robustness of the proposed system for air bubble disturbance on the image, aiming at confirming the control characteristics of the proposed visual servoing system. The following results were derived; (1) The proposed system is robust to time-variant target position in z-axis (front-back direction of the vehicle). (2) Although the fitness value of GA is influenced by disturbance, the system can keep recognizing the position and orientation of 3D marker, and tracking by visual servoing could be maintained under the air bubble disturbances. Key Words: AUV, ROV, Visual servoing, Genetic algorithm ͍ͯɼແࡧԽɾࣗԽ͞ΕͨਫதϩϘοτͰ͋Δ AUV ͷ 1. ͡ Ί ʹ ։ൃ͕ॏཁͰ͋Δɽ͔͠͠ɼ࡞ۀରΛ͍͔ʹը૾ೝࣝ͞ ؾதӉͰ͢༻׆ΔͨΊͷϩϘοτͷڀݚ։ൃͷݱঢ়ͱ ͤΔ͔ͱ͍͏͕͞Ε͍ͯΔɽ ൺֱͯ͠ɼਫதϩϘοτͦͷ༻׆ͷՄೳੑͷେ͖͞ʹؔΘ ਫதϩϘοτʹؔ͢Δදతͳը૾ೝࣝख๏ͱͯ͠ɼจ Βͣɼ͍·ͩൃల్্ͷஈ֊Ͱ͋Δɽಛʹ࡞ۀରΛ͍͔ ݙ1), 2) ͰਫதϩϘοτʹରΛೝࣝͤ͞ΔͨΊɼڑ ʹը૾ೝࣝͤ͞Δ͔͕ɼ༻׆ͷՄೳੑΛ֦͛ΔͨΊͷΩʔϙΠ ใΛ୯؟ΧϝϥͰऔಘ͢Δํ๏͕հ͞Ε͍ͯΔɽϚϦϯε ϯτͰ͋Δͱߟ͑ΒΕΔɽਫதϩϘοτ ROV(Remotely ϊʔͳͲ͕औಘը૾ʹөΓࠐΉͱɼରܗঢ়ͷಛͷҰͭ Operated Vehicle) AUV(Autonomous Utility Vehicle) Ͱ͋ΔΤοδͷݕग़͕ࠔʹͳΔ͕ɼΧϥʔநग़ॲཧʹΑͬ ʹେผͰ͖ΔɽROV Ͱͷ࡞ۀɼέʔϒϧΛհͯ͠ૹΒΕ ͯ֎ཚͷআͱڈରͷநग़Λಉ͔࣌ͭϦΞϧλΠϜʹߦ͏ ͯ͘ΔΧϝϥը૾ͷใʹ ͕ऀ࡞ૢ͖ͮجROV ͷԕִૢ ख๏͕ఏҊ͞Ε͍ͯΔɽ͞ΒʹɼਫதϥϯυϚʔΫͱͯ͠Ґ ࡞Λߦ͍ͬͯΔ͕ɼ࡞ۀͷਫਂ͕ਂ͘ͳΔʹͭΕɼ࡞ྖۀҬ ஔඪఆඪ͕ઃ͞ܭΕ͓ͯΓɼͦͷطʹΧϥʔϘʔϧΛ ʹ౸ୡ͢Δ·Ͱͷ࣌ؒճऩʹ͕͔͔࣌ؒͬͯ͠·͏ɽ͞Β ༻͢Δͱͱʹલड़ͷख๏ͱΈ߹Θͤͯը૾্Ͱൃ͠ݟ ʹɼਫਂʹԠͯ͡έʔϒϧ͘ͳΔͨΊɼέʔϒϧΛؚ ͍͢Λߦ͍ͬͯΔɽͳ͓ɼ͜ͷख๏Ͱط͕ӅΕ ΊͨγεςϜશମେ͖͘ͳͬͯ͠·͏ɽͦͷͨΊɼൣғ ͯൃ͍࣌ͳ͖ͰݟͷਫதϩϘοτͷ࢟ͷํࢉܭ๏ʹ͍ͭͯ ͔ͭւఈܗঢ়ະͳ͚͓ʹڥΔւఈࢿݯͷ୳ࡧͳͲʹ͓ ड़ΒΕ͍ͯΔ͕ɼ୯ ؟2 ࣍ݩը૾ΛͱʹਫதϩϘοτ ͷҐஔɾ࢟ͷࢉܭΛߦ͍ͬͯΔɽ ԬࢁେֶେֶӃࣗવՊֶڀݚՊ ∗ Graduate School of Natural Science and Technology, Okayama University ʢReceived September 8, 2015ʣ ∗ ͜Εʹର͠ஶऀΒ͕ఏҊ͍ͯ͠Δ࣮࣌ؒҨతೝࣝख๏ 3), 4) ɼࡾͭͷΧϥʔ͔ٿΒͳΔରͷ 3 ࣍ݩϞσϧͱͯ͠ 3 ࣍ݩϚʔΧΛਫதϩϘοτʹఆٛ͠ɼ3 ࣍ݩϚʔΧͷҐஔɾ c 2015 SICE TR 00xx/xx/xxxx–0001 ° T. SICE Vol.xx No.xx Xxx 20xx 2 Fig. 1 Overview of ROV (a)Front view (b)Side view (c)Top view (d)Back view Fig. 2 Layout of under water experimental devices NTSCɼ࠷ඃࣸମর 0.8[lx]ɼζʔϜͳ͠ʣΛࡌ͓ͯ͠ ࢟Λ GA ʹΑͬͯΦϯϥΠϯͰͤ͞ࢉܭɼద߹ͷ࠷ߴ Γɼରͷ 3 ࣍ݩೝࣝΛߦ͏ͨΊɼલํ 2 ͷΧϝϥΛಉ ͍Ҩࢠ͕࣋ͭใΛਫதϩϘοτ͕ೝࣝͨ͠ 3 ࣍ݩϚʔΧ ࣌༻ͨ͠ɽਫதಈྗͰܥɼਫฏεϥελ 2 جʢ࠷େਪྗ ͷҐஔɾ࢟ͱͯ͠ࢉग़͍ͯ͠Δɽ͜ͷద߹ͷ࣌ม 9.8[N]ɽҎԼಉ༷ʣɼਨεϥελ 1 جʢ4.9[N]ʣɼԣεϥελ ͰଟๆੑͱͳΔ͕ɼ࣮࣌ؒͷ੍ޚϧʔϓ͝ͱʹΦϯϥΠϯͰ 1 جʢ4.9[N]ʣΛࡌ͍ͯ͠Δɽ·ͨɼর֬อͷͨΊ LED ద߹ͷ࠷ߴ͍ҨࢠΛೝࣝ݁Ռͱͯ͠બͿΑ͏ʹ͍ͯ͠ ϥΠτʢ5.8WʣΛ 2 جࡌ͍ͯ͠Δɽ ΔͨΊɼͨͱ͑Ұ࣌తʹద߹͕Լ͕ͬͨͱͯ͠ɼ੍ޚϧʔ ͭ͗ʹɼਫಓਫΛຬͨͨ͠؆қϓʔϧʢॎʷԣʷߴ͞ɼ2[m] ϓ͕܁Γฦ͞ΕΔ͜ͱͰద߹Λৗʹߴ͘ҡ࣋͠Α͏ͱ͢Δ ʷ 3[m] ʷ 0.75[m]ʣΛ࣮ݧ૧ͱͯ͠༻ͨ͠ɽ্ هROV Λ ੑ࣭Λ͍࣋ͬͯΔɽଞͷख๏ͱҟͳΔ 3 ࣍ݩϚʔΧΛෳ ࣮ݧ૧ʹೖਫͤ͞ɼਤ 2 ʹࣔ͢Α͏ʹ ROV ͷΧϝϥը૾ͷ Ͱ؟ೝ͍ࣝͯ͠ΔͰ͋Γɼࠨӈը૾໘ʹࣸӨ͠ɼ࣮ը૾ͱ ड৴ɼPC ͔Βͷ੍ޚ৴߸ͷૹ৴ͳΒͼʹిྗڅڙςβʔ ͷ૬ؔΛ GA ਐԽͷద߹ͱͯ͠༻͍Δ͜ͱͰ 3 ࣍ݩର έʔϒϧʢ200[m]ʣΛհͯ͠ߦͬͨɽPC ଆͰ ROV ࡌ ͷҐஔɾ࢟Λ࣮࣌ؒͰܭଌ͍ͯ͠Δ 4) ɽ ຊͰڀݚ͜Ε·ͰਫதϩϘοτͷਫதࣗಈॆిΛࢦ͠ ͨࣗಈቕ߹੍ݧ࣮ޚΛߦ͍ͬͯΔ͕ 5) ɼೝࣝը૾ʹ·ؚΕΔ ͷ 2 ͷΧϝϥ͔ΒૹΒΕΔରͷը૾ใΛ ʹجGA ʹ ΑΔϞσϧϕʔετϚονϯάΛߦͬͨɽ 2. 2 ਫத࣮ݧ݅ ֎ཚʹؔͯ͠ରͷҐஔɾ࢟ͷೝࣝਫ਼ͷ੍ޚಛੑͷݕ 2. 2. 1 ਫதʹ͓͚ΔҐஔɾ੍࢟ݧ࣮ޚ ౼ߦ͍ͬͯͳ͔ͬͨɽ࣮ւҬʹ͓͍ͯਫதϩϘοτͱೝ ਫதʹઃஔͨ͠ 3 ࣍ݩϚʔΧʢϘοΫεʢ100[mm] × ࣝରؒʹϚϦϯεϊʔνϜχʔ͔Βग़͢Δਫ͕ଘ 100[mm] × 100[mm]ʣपΓʹɾ੨ɾͷΧϥʔٿʢܘ ࡏ͢Δ߹͋Δͱߟ͑ΒΕΔ͕ɼ͜ΕΒ͕ਫதϩϘοτͷ 40[mm]ʣΛஔͨ͠ରʣΛ GA ʹΑΓೝࣝͤ͞ɼਫதϩ औಘը૾ͷதͰରΛःΔΑ͏ʹөΓࠐΉͱɼਖ਼֬ͳೝࣝ Ϙοτʹࡌͨ͠ 4 جͷεϥελࢦྩిѹΛૹΔ͜ͱͰɼ ݁Ռ͕ಘΒΕͣҐஔɾ੍͕࢟ޚཚ͞ΕΔՄೳੑ͕͋Δɽ͢ ਫதϩϘοτͱ 3 ࣍ݩϚʔΧͷؒͰҎԼͷ૬ରతඪҐஔɾ ͳΘ࣮ͪւҬͰͷ༻Λߟ͑ͨ߹ɼରΛःΔΑ͏ʹө ࢟ (xd [mm]ɼyd [mm]ɼzd [mm]ɼǫ2d [deg]) ΛอͭΑ͏ʹϨ ΓࠐΉ֎ཚʹର͠ɼ։ൃதͷਫதϩϘοτ͕ͲͷఔϨΪϡ ΪϡϨʔτ੍ͤͨ݁͞ޚՌͷূݕΛߦͬͨɽͳ͓ɼGA ʹΑ ϨʔτੑೳΛൃ͖ͰشΔ͔֬ೝ͢Δ͜ͱ͕ॏཁͳͰ͋Δɽ ΔϞσϧϕʔετϚονϯά͔Βࢉग़͞ΕͨରͷҐஔɾ ͦ͜ͰຊͰڀݚւதͰͷը૾ೝࣝʹର͢Δ༷ʑͳ֎ཚΛؾ ๐ʹΑͬͯγϛϡϨʔτ͠ɼ࣮࣌ؒҨతೝࣝख๏ 3) ͷϩό ࢟ͷೝࣝใɼਫதϩϘοτͱରؒͰઃఆͨ͠ GA ୳ࡧྖҬʢΣH Λத৺ͱͯ͠ xH ࣠ํʹ ±400[mm]ɼyH ετੑΛ֬ೝ͢Δͱͱʹɼؾ๐֎ཚதͰϏδϡΞϧαʔ ࣠ํʹ ±200[mm]ɼzH ࣠ํʹ +800[mm] ͷྖҬʣͰ֫ Ϙ͕ՄೳͰ͋Δ͜ͱΛࣔ͢ɽ ಘ͞ΕΔͱԾఆ͍ͯ͠Δɽਤ 3 ʹຊڀݚͷ࣮͚͓ʹݧΔ࠲ඪ 2. ࣮ ํ ݧ๏ 2. 1 ਫத࣮ڥݧ จ ݙ5) ͱಉ͡ਫத࣮ڥݧΛ༻ҙͨ͠ɽҎԼʹུ֓Λड़ ܥΛࣔ͢ɽ͜͜Ͱ xd = H xM = 0, yd = H yM = −67, zd = H zM = 600, ǫ2d = 0 Δɽ·ͣԕִૢ࡞ܕਫதϩϘοτʢ ( )ג ROVɼ࠷େਫ Ͱ͋Δɽ·ͨɼx[mm]ɼy[mm]ɼz[mm]ɼǫ2 [deg] Λ GA Ͱೝ ਂ 50mʣΛਤ 1 ʹࣔ͢ɽຊϩϘοτෳ֮ࢹ؟ηϯαʔͱ͠ ࣝͨ͠ਫதϚʔΧʔͷҐஔɾ࢟ͱ͓͖ɼ૬ରతඪҐஔɾ ͯɼԕִૢॎ༻νϧτ͖ߏػΧϝϥ 1 ʢࡱ૾ૉࢠ CCDɼ ࢟ʹਫதϩϘοτΛϨΪϡϨʔτͤ͞ΔͨΊɼҎԼͷ P ੍ ըૉ 38 ສըૉɼ৴߸ํࣜ NTSCɼ࠷ඃࣸମর 1.5[lx]ɼ ͔ޚΒ͞ࢉܭΕΔࢦྩిѹ v1 ∼ v4 Λ֤εϥελ༩͑ͨɽ ޫֶζʔϜ 10 ഒʣͱ੍ͯ͠ͱ༻ޚલํ͓ΑͼԼํʹͦΕͧ ͳ͓ ǫ2 ਤ 3 ʹࣔ͢ ΣH ͷ y ࣠ճΓͷ֯Ͱ͋Δɽx ͓࣠ ΕΧϝϥ 2 ʢࡱ૾ૉࢠ CCDɼըૉ 38 ສըૉɼ৴߸ํࣜ Αͼ z ࣠ճΓͷ֯ਫதϩϘοτͷߏʹىҼʢු৺͕ॏ ܭଌࣗಈ੍ֶޚձจू ୈ xx רୈ xx ߸ 20xx xx ݄ 3 yd = −67[mm]ɼzd = 341[mm]ɼǫ2d = 0[deg] ͱͳΔΑ͏ʹ ઃஔͨ͠ɽ·ͨɼਫதϩϘοτͷΧϝϥը૾ʹө͠ग़͞Εͨ 3 ࣍ݩϚʔΧલํʹɼ࣮ݧ։࢝ 30[s] ͔ޙΒؾ๐֎ཚ͕өΓࠐ ΉΑ͏ͳঢ়گΛൃੜͤͨ͞ɽͦͷ্Ͱɼ3 ࣍ݩϚʔΧͷഎܠ ͱ࣮ͯ͠ݧਫ૧͕ͦͷ··өΓࠐΉɼ͢ͳΘͪഎ͠ͳܠͷ ߹ͱ࣮ڥΛٖͨ͠ւதͷࣸਅΛషΓ͚ͨʢഎ͋ܠΓͷʣ ߹ʹ͍ͭͯɼద߹ͷߴ͍ॱʹ্Ґ 36 ݸͷҨࢠ͕࣋ͭ Fig. 3 Coordinate system provided in under water experiment ೝࣝ݁Ռʢೝࣝͨ͠ 3 ࣍ݩϚʔΧͷҐஔɾ࢟ʣΛௐͨɽ 3 ࣍ݩϚʔΧʹରͯ͠എ͠ͳܠͷ߹ͷ݁ՌΛਤ 4ɼਤ 5 ʹ ࣔ͠ɼഎ͋ܠΓͷ߹ͷ݁ՌΛਤ 6ɼਤ 7 ʹࣔ͢ɽਤ 4ɼ6 ৺ͷ্ʹҐஔ͍ͯ͠Δ͜ͱʣ͢Δ෮ݩτϧΫʹΑΓɼ࣌ఆ ͦΕͧΕ (a) ೝࣝͨ͠ x ࣠ํͷ 3 ࣍ݩϚʔΧͷҐஔ (b)y 1 ඵఔͰࣗతʹ҆ఆঢ়ଶʹ෮͢ݩΔɽͦͷͨΊຊͰڀݚ ࣠ํͷҐஔ (c)z ࣠ํͷҐஔ (d)y ࣠पΓͷ֯ (e) ࣮ݧ x ͓࣠Αͼ z ࣠ճΓͷ֯ͷ੍ޚߦΘͳ͍ɽ ։࢝ 10 ඵޙʢؾ๐֎ཚͳ͠ʣͷ 3 ࣍ݩϚʔΧͷೝࣝ݁Ռ (f) Ԟߦ͖ํ : v1 = kp1 (zd − z) + 2.5 ʢv1 = 0[V] ͷͱ͖ ΣH ͷ zH ࣠ํʹ ਪྗ 9.8[N]ɼv1 = 5[V] ͷͱ͖ −9.8[N]ʣ Ԗ࣠ճస : v2 = kp2 (ǫ2d − ǫ2 ) + 2.5 ʢv2 = 0[V] ͷͱ͖ ΣH ͷ yH ࣠ճΓʹ ટճτϧΫ 0.88[Nm]ɼv2 = 5[V] ͷͱ͖ −0.88[Nm]ʣ ࣮ݧ։࢝ 40 ඵޙʢؾ๐֎ཚ͋Γʣͷ 3 ࣍ݩϚʔΧͷೝࣝ݁Ռ Λද͍ͯ͠Δɽ͜͜Ͱ 3 ࣍ݩϚʔΧͷೝࣝ݁Ռͱɼਤ 4ɼ6 ͷ (e) (f) ʹ͓͍ͯ 3 ࣍ݩϚʔΧʹઃஔ͞Εͨɼ੨ɼͷ ۙٿͷ֤৭ͷઢͷԁͰද͞Ε͓ͯΓɼຊγεςϜͰ͜ ͷೝࣝΛ࣮࣌ؒʢ33[ms]ʣͰ܁Γฦ͠ߦ͍ͬͯΔɽΧϝϥը ૾্ʹ͓͍ͯɼ֤৭ͷઢͷԁ͕ϚʔΧʹઃஔ͞Ε֤ͨٿͷ ֎पͱશʹҰக͍ͯ͠Δ߹ɼҨࢠͷద߹͕࠷ߴ͘ɼ ϚʔΧͱͷ૬ରతҐஔɾ͕࢟ਖ਼͘͠ೝࣝ͞Ε͍ͯΔ͜ͱΛ ҙຯ͍ͯ͠Δɽਤ 4(a)∼(d) ΑΓɼҐஔɾ࢟ͷೝࣝ݁Ռ͕࣮ Ԗํɹ : v3 = kp3 (yd − y) + 2.5 ݧ։࢝ʹޙऩଋ͍͕ͯ͘͠ɼؾ๐֎ཚൃੜޙൃੜલʹൺ ʢv3 = 0[V] ͷͱ͖ ΣH ͷ yH ࣠ํʹ ͯҨࢠͷೝࣝ݁Ռ͕Βͭ͘͜ͱΛ֬ೝͰ͖Δɽ͜ΕΧ ਪྗ −4.9[N]ɼv3 = 5[V] ͷͱ͖ਪྗ ϝϥը૾ʹөΓࠐΜͩؾ๐֎ཚʹΑͬͯɼ3 ࣍ݩϚʔΧͷҐ ஔͱ࢟ͷೝ͕ࣝ͘͠ͳͬͨ͜ͱΛҙຯ͍ͯ͠Δɽ͞Βʹ 4.9[N]ʣ ਫฏํɹ : v4 = 5.0 ʢxd − x < −5[mm] ͷͱ͖ ΣH ͷ xH ࣠ํʹਪྗ −4.9[N]ʣ 0.0 ʢxd − x > 5[mm] ͷͱ͖ ΣH ͷ xH ࣠ํʹਪྗ 4.9[N]ʣ ͞ΒʹຊใͰɼೝࣝը૾ʹөΓࠐΉ֎ཚʹର͢Δϩόετ ੑΛ͢ূݕΔͨΊɼϨΪϡϨʔτதͷਫதϩϘοτʹରͯ͠ɼ νϜχʔ͔Βग़͢Δؾ๐Λٖͨ͠ڥΛਤ 3 ʹࣔ͢ 3 ࣍ ࣮ڥΛٖͨ͠എ͋ܠΓͷ߹ͷ݁Ռʢਤ 6(a)∼(d)ʣɼ എ͠ͳܠͷ߹ͱൺֱͯ͠ೝࣝ݁Ռ͕Βͭ͘ͷͷɼഎܠ ͳ͠ͷ߹ͱಉ༷ʹϚʔΧͱͷ૬ରతҐஔɾ࢟Λೝࣝ͠ଓ ͚ΒΕΔ͜ͱ͕֬ೝͰ͖Δɽͳ͓ɼؾ๐֎ཚͳ͠ͷঢ়ଶʢ࣮ ݧ։࢝∼ ޙ30[s] ·ͰʣͰਤ 4(c) (a) ͓Αͼ (b) ͱൺ ͯೝࣝ݁Ռ͕Β͍͍ͭͯΔɽ͜ΕΧϝϥը૾ͷ্Լࠨ ӈʢx ͓Αͼ y ࣠ʣํͱൺͯɼԞߦ͖ʢz ࣠ʣํͷҐ ஔͷೝ͕͍ࣝ͜͠ͱΛද͍ͯ͠Δɽ ͭ͗ʹਤ 5ɼ7 ͦΕͧΕద߹ͷΛද͍ͯ͠Δɽਤ தͷࠇ৭ͷ͕֤Ҩࢠͷೝࣝ݁Ռʢద߹ͱͦͷ࣌ͷೝࣝ ݩϚʔΧͷલํʹஔ͠ɼग़͢Δؾ๐ͷ༗ແʹΑͬͯඪҐ ҐஔʣͰ͋Δɽ۩ମతʹ֤ਤʹ͓͍ͯ (a) ࣮ݧ։࢝ ޙ10[s] ஔɾ࢟ͷ෮ݩೳྗ͕ͲͷΑ͏ʹมԽ͢Δ͔ͷ֬ೝΛߦ͏ɽ ʢؾ๐֎ཚͳ͠ʣʹ͓͚Δద߹ͱͦΕʹରԠ͢Δ x ͓Αͼ y 3. ݁Ռͱߟ ࠲ඪͷೝࣝҐஔɼ(b) ࣮ݧ։࢝ ޙ40[s]ʢؾ๐֎ཚ͋Γʣʹ͓ ͚Δ x ͓Αͼ y ࠲ඪͷೝࣝҐஔɼ(c) ࣮ݧ։࢝ ޙ10[s] ʹ͓ 3. 1 GA ʹΑΔ 3 ࣍ݩϚʔΧͷೝࣝੑೳ ͚Δ x ͓Αͼ z ࠲ඪͷೝࣝҐஔɼ(d) ࣮ݧ։࢝ ޙ40[s] ʹ͓ ຊઅͰɼఏҊγεςϜͰ༻͍ͨ GA ʹΑΔ 3 ࣍ݩϚʔ ͚Δ x ͓Αͼ z ࠲ඪͷೝࣝҐஔͰ͋Γɼ(e)∼(h) (a)∼(d) ΧͷҐஔɾ࢟ͷೝ࣮ࣝ݁ݧՌʹ͍ͭͯࣔ͠ɼఏҊख๏ͷϩ ͷਤʹ͓͍ͯద߹ͷͷࢁͷۙͷྖҬΛ֦େͨ͠ όετੑʹ͍ͭͯߟ͢Δɽຊ࣮Ͱݧ GA ͷ݅ͱͯ͠Ҩ ਤͰ͋ΔɽͦΕͧΕͷਤΑΓɼഎܠͷ༗ແʹ͔͔ΘΒͣɼؾ ࢠΛ 60ɼදܕݱΛ 3 ࣍ݩϚʔΧͷҐஔͱ࢟ʢx[mm]ɼ ๐֎ཚͳ͠ͷ߹ͱൺֱͯ͠ؾ๐֎ཚ͕ଘࡏ͢Δ߹ద߹ y[mm]ɼz[mm]ɼǫ2 [deg]ʣɼ੍ޚपظʢ1 ճͷೖྗը૾ʹର͠ ͕͘ͳΔɽ͔͠͠ɼద߹͕࠷ߴ͍ҨࢠͷೝࣝҐஔ ͯҨࢠΛਐԽͤ͞Δ࣌ؒʣΛ 33[ms]ɼ੍ޚपظຖͷੈͷ ؾ๐֎ཚ࣮ڥΛٖͨ͠എܠͷ༗ແʹ͔͔ΘΒͣ΄ͱ ߋ৽ճΛ 9 ճͱ͍ͯ͠Δɽ·ͨɼ3 ࣍ݩϚʔΧਫதͰݻఆ ΜͲมԽ͕ແ͍͜ͱ͕͔ͬͨɽҎ্ΑΓɼఏҊγεςϜͷ ͠ɼਫதϩϘοτͱͷ૬ରతҐஔɾ࢟ͷਅ͕ xd = 0[mm]ɼ GA 3 ࣍ݩϚʔΧͷҐஔɾ࢟Λ࣮࣌ؒͰೝࣝͰ͖Δϩό 4 T. SICE Vol.xx No.xx Xxx 20xx ετੑΛ༗͍ͯ͠Δ͜ͱ͕֬ೝͰ͖Δɽ Fig. 4 Distribution of top 36 genes in case of plain background: (a)Position in x-axis direction, (b)Position in y-axis direction, (c)Position in z-axis direction, (d)Angle around y-axis, (e)Left and right camera images at 10[s] and (f)Left and right camera images at 40[s]. 3. 2 ਫதʹ͓͚Δ GA ೝࣝਫ਼ͱϨΪϡϨʔτੑೳ ਤ 8(a) 3 ࣍ݩϚʔΧલʹؾ๐Λग़ݱʢਤ 3 ʹࣔ͢ʣ͞ ͤɼೝࣝը૾ʹ֎ཚΛөΓࠐ·ͤͨঢ়ଶͰ૬ରతඪҐஔɾ ࢟ xd = 0ɼyd = −67ɼzd = 600ɼǫ2d = 0 ʹϨΪϡϨʔ τͤͨ͞ਫதϩϘοτͷ GA ೝࣝ࣌ͷద߹ͷ࣌ؒมԽΛࣔ ͢ɽ͜ͷਤͷ (a) ʹΑΔͱɼೝࣝ։͔࢝ΒඵҎͰద߹ ͕ 0.4 ∼ 1 ͷؒΛਪҠ͠ɼฏݧ࣮ͯ͠ۉத 0.7 લޙͷద߹ Λҡ͍࣋ͯ͠Δ͜ͱ͕Θ͔Δɽਤ 8(b)ʙ(e)ɼ(g)ʙ(j) ૬ ରతඪҐஔɾ࢟ͱ GA Ͱೝࣝͨ͠ਫதϚʔΧͷҐஔɾ࢟ Fig. 5 Fitness distribution generated by each gene in case of plain background: (a)Fitness distribution between x and y positions at 10[s], (b)Fitness distribution between x and y positions at 40[s], (c)Fitness distribution between x and z positions at 10[s], (d)Fitness distribution between x and z positions at 40[s], (e)Enlarged view of (a), (f)Enlarged view of (b), (g)Enlarged view of (c) and (h)Enlarged view of (d). ͱͷࠩޡɼ͓ΑͼͦΕΒΛ෮͢ݩΔྗɾτϧΫΛද͍ͯ͠ Δɽ·ͨɼ(f) ϨΪϡϨʔτதͷਫதϩϘοτͷҐஔΛਤ 3 ͷਫதϩϘοτͱ 3 ࣍ݩϚʔΧͷҐஔؔʢࠨྻʣɼਫதϩ தͷ ΣH Λج४ʹද͍ͯ͠ΔɽGA ͷೝࣝࠩޡɼϩϘοτҠ Ϙοτ͔Β ͨݟ3 ࣍ݩϚʔΧͷҐஔʢӈྻʣΛ 10[s] ຖʹऔ ಈ࣌ʹੜ͡Δςβʔέʔϒϧ͔Βͷྗɼ͞ΒʹҠಈ࣌ʹൃ ಘͨ͠ը૾Λ͍ࣔͯ͠Δɽਤ 9(a)ʙ(e) ͷӈྻͷ֤ը૾ʹ͓ ੜ͢ΔਫѹมԽʹΑΔ࣮ݧ૧ଆ໘͔ΒͷࣹΛड͚ɼ૬ର ͍ͯɼ3 ࣍ݩϚʔΧͷ֤΅΄ͱٿಉ͡Ґஔʹඳը͞Ε͍ͯΔ తඪҐஔɾ͔࢟Βͷ͕ࠩޡఆৗతʹݱΕΔͷͷɼ4 جͷ ɼɼ੨ͷԁ GA ͕࣮࣌ؒͰೝࣝͨ͠ 3 ࣍ݩϚʔΧͷҐ εϥελΛ੍͢ޚΔ͜ͱͰ֎ཚཁૉΛΩϟϯηϧ͠ɼ૬ରత ஔɾ࢟Λද͍ͯ͠Δɽ͜ΕΒͷඳը͞Εͨԁͱ 3 ࣍ݩϚʔ ඪҐஔɾ࢟ۙʹϨΪϡϨʔτͰ͖Δ͜ͱ͕֬ೝͰ͖Δɽ Χͷ֤ٿͷେ͖͕ͯ͢͞Ұக͍ͯ͠Δ߹ɼຊγεςϜ 3 ࣍ݩϚʔΧͷҐஔɾ࢟Λ͘ͳࠩޡೝ͍ࣝͯ͠Δʢద߹ 3. 3 3 ࣍ݩϚʔΧ͕पظӡಈ͢Δ߹ ͕࠷ߴ͍ʣ͜ͱΛද͍ͯ͠Δɽ͜ΕΒͷਤ (a)ʙ(e) ͷӈྻ ૬ରతඪҐஔɾ࢟Λ 3.1 ͱಉ͡ͱͯ͠ 3 ࣍ݩϚʔΧ ͷ֤ը૾ʹඳը͞Εͨɼɼ੨ͷԁͷҐஔ͔Β͔ΔΑ͏ લํʹؾ๐Λग़ ͭͭͤ͞ݱ3 ࣍ݩϚʔΧΛपظӡಈͤͨ͞ ʹɼೝࣝը૾ʹ֎ཚʢؾ๐ʣ͕өΓࠐΜͰ͍Δʹ͔͔ΘΒ ߹ͷਫதϩϘοτͷϨΪϡϨʔτੑೳʹ͍ͭͯ֬ೝΛߦ͏ɽ ͣɼϚʔΧͱਫதϩϘοτͷ૬ରతҐஔҰఆʹอͨΕ͍ͯ 3 ࣍ݩϚʔΧલํʹؾ๐Λग़ͨͤ͞ݱ߹ʹ͍ͭͯͷ࣮݁ݧ Δ͜ͱ͕͔Δɽ͞Βʹ (a)ʙ(e) ͷӈଆࣸਅதʹɼGA ͕ ՌΛਤ 9 ͱਤ 10 ʹࣔ͢ɽਤ 9 3 ࣍ݩϚʔΧΛਤ 3 ʹࣔ͢ ࣮࣌ؒͰೝࣝͨ͠ 3 ࣍ݩϚʔΧͷҐஔɾ͕࢟ɼೝࣝը૾த ΣH ͷ z ࣠ํʹप ظ20[s]ɼৼ෯ 280[mm] Ͱӡಈͤͨ࣌͞ ͷ࣮ࡍͷ 3 ࣍ݩϚʔΧͷҐஔɾ࢟ͱ͓͓ΑͦҰகͯࣔ͠͞ ܭଌࣗಈ੍ֶޚձจू ୈ xx רୈ xx ߸ 20xx xx ݄ 5 Fig. 6 Distribution of top 36 genes in case of simulated ocean background: (a)Position in x-axis direction, (b)Position in y-axis direction, (c)Position in z-axis direction, (d)Angle around y-axis, (e)Left and right camera images at 10[s] and (f)Left and right camera images at 40[s]. Ε͓ͯΓɼ࣮࣌ؒ 3 ࣍ݩೝ͕ࣝҡ࣋͞Ε͍ͯΔ͜ͱ͕͔Δɽ ͍͑ݴΕɼຊγεςϜೝࣝը૾தʹؾ๐ʹΑΔ֎ཚ͕ өΓࠐΜͩͱͯ͠ɼ࣮࣌ؒͰ࿈ଓతʹ 3 ࣍ݩϚʔΧͷҐஔɾ ࢟Λ GA ͕ೝࣝ͠ଓ͚Δ͜ͱ͕ՄೳͰ͋ΔͨΊɼͦͷ݁Ռ ͱͯ͠૬ରతඪҐஔɾ࢟ͷϨΪϡϨʔτੑೳߴ͍͜ ͱΛ͍ࣔࠦͯ͠Δɽ·ͨɼ͜ͷ࣮݁ݧՌΛਤ 10 ʹࣔ͢ɽਤ 10(a)ʙ(d) ɼͦΕͧΕద߹ɼz ࣠ํͷਫதϩϘοτͷ Ґஔɼ૬ରతඪҐஔͱͷࠩޡɼࠩޡΛ෮͢ݩΔྗͷάϥϑ Λ͍ࣔͯ͠Δɽ·ͨɼ(a)(b) தʹ( ͨ͠ࡌهA) ͓Αͼ (B) ͱ ͦΕͧΕͷӈͷҹɼ࣮ݧ։࢝ 20[s] ͔ޙΒલड़ͷपৼͱظ Fig. 7 Fitness distribution generated by each gene in case of simulated ocean background: (a)Fitness distribution between x and y positions at 10[s], (b)Fitness distribution between x and y positions at 40[s], (c)Fitness distribution between x and z positions at 10[s], (d)Fitness distribution between x and z positions at 40[s], (e)Enlarged view of (a), (f)Enlarged view of (b), (g)Enlarged view of (c) and (h)Enlarged view of (d). ෯Ͱ 3 ࣍ݩϚʔΧΛӡಈͤͨؒ͞ظɼ࣮ݧ։࢝ 10[s] ͔ޙΒ ೝࣝը૾தʹؾ๐Λग़ؒظͨͤ͞ݱΛද͍ͯ͠Δʢͳ͓ɼ(b) 4. · ͱ Ί ͷਤͷҰઢਫதϩϘοτͷඪيಓΛද͍ͯ͠Δʣ ɽਤ 10(a) ͕ࣔ͢Α͏ʹؾ๐ͷग़ʹݱΑͬͯద߹͕Լ͢Δ͕ɼ ROV Λ AUV Խ͢ΔͨΊͷڀݚ։ൃΛਐΊΔͨΊɼຊڀݚ ਤ 10(b)(c) ΑΓɼ૬ରతඪҐஔʹϨΪϡϨʔτ͍ͯ͠Δ༷ Ͱ 3 ࣍ݩϚʔΧΛ༻͍ͨϏδϡΞϧαʔϘܕਫதϩϘοτ ࢠ͕֬ೝͰ͖Δɽ͢ͳΘͪɼఏҊ͢ΔγεςϜೝࣝը૾த ͷؾ๐֎ཚʹର͢ΔҐஔɾ੍࢟ޚͷݧ࣮ূݕΛߦ͍ɼԼ݁ه ʹ֎ཚ͕өΓࠐΜͩͱͯ͠ɼ૬ରతඪҐஔʹϨΪϡϨʔ Λಘͨɽ(1) ఏҊγεςϜ z ࣠ํͷ࣌มඪʹର͠ τͰ͖ΔೳྗΛ༗͢Δ͜ͱ͕ࣔ͞Εͨɽ ͯϩόετʹैͰ͖Δɽ(2) ؾ๐֎ཚ͕өΓࠐΉڥԼͰ ͞Βʹɼؾ๐Λग़ͨͤ͞ݱঢ়ଶͰ 3 ࣍ݩϚʔΧͷӡಈपظ GA ͷద߹͕Լ͢Δ͕ɼఏҊγεςϜ 3 ࣍ݩϚʔΧ Λप ظ15[s]ɼৼ෯ 280[mm] ʹͨ࣌͠ͷ݁ՌΛਤ 11 ʹࣔ͠ɼ ͱͷ૬ରతҐஔɾ࢟Λϩόετʹೝࣝ͠ଓ͚Δ͜ͱ͕Ͱ͖ 3 ࣍ݩϚʔΧͷӡಈपظΛप ظ10[s]ɼৼ෯ 280[mm] ʹͨ͠ Δɽ͞Βʹɼؾ๐֎ཚԼʹ͓͍ͯϏδϡΞϧαʔϘʹΑΔ ࣌ͷ݁ՌΛਤ 12 ʹࣔ͢ɽ͍ͣΕͷ݅Ͱɼೝࣝը૾தʹؾ ࣌มඪͷै੍ޚΛϩόετʹߦ͏͜ͱ͕Ͱ͖Δɽࠓ ๐͕өΓࠐΉͱద߹ͷ͕͘ͳΔ͕ɼ૬ରతඪҐஔʹ ޙਫதࣗಈॆిʹ͚ͨϏδϡΞϧαʔϘܕਫதϩϘοτ ϨΪϡϨʔτͰ͖Δ͜ͱΛ֬ೝͰ͖Δɽ͢ͳΘͪɼؾ๐֎ཚ ͷيಓै੍ޚ๏ͷݕ౼Λߦ͏ɽ ʹରͯ͠ఏҊ͢ΔγεςϜ૬ରతඪҐஔͷϨΪϡϨʔ τੑೳ͕ϩόετͰ͋Δ͜ͱ͕͔ͬͨɽ ँࣙ ຊͰڀݚࣜגձࣾϚϦϯγεςϜ෦ͷྗڠΛ ಘ·ͨ͠ɽ͜͜ʹँҙΛද͠·͢ɽ T. SICE Vol.xx No.xx Xxx 20xx 6 Fig. 8 Regulator performance with additional disturbance made by air bubbles against dual-eye image recognition: (a)fitness value, (b)error in x-axis direction, (c)error in y-axis direction, (d)error in z-axis direction, (e)error around y-axis, (f)3D trajectory of underwater vehicle (g)thrust in x-axis direction, (h)thrust in y-axis direction, (i)thrust in z-axis direction and (j)torque around y-axis ࢀ ߟ จ Fig. 9 Actual snapshot of underwater vehicle (left column) and its camera images (right column) with disturbance on images after starting experiment: (a)10 seconds have passed, (b)20 seconds have passed, (c)30 seconds have passed, (d)40 seconds have passed and (e)50 seconds have passed ݙ 1ʣ༄ળమ, Ӝ, ౻Ҫً, ਫதϥϯυϚʔΫΛར༻ͨ͠ਫதϩ Ϙοτͷߤ๏੍ޚ, ੜ࢈ڀݚ, 52 ר, 5 ߸ (2000), pp.247-250. 2ʣ༄ળమ, Ӝ, ౻Ҫً, ۙ౻ҳਓ, ਓਫதϥϯυϚʔΫͱਪ ଌߤ๏Λར༻ͨࣗ͠ܕਫதϩϘοτͷߤ๏, ຊϩϘοτֶ ձࢽ, Vol.20, No.3 (2002), pp.290-298. 3ʣSuzuki, H. and Minami, M., Visual Servoing to catch fish Using Global/local GA Search, IEEE/ASME Transactions on Mechatronics, Vol.10, No.3 (2005), pp.352-357. 4ʣNishimura, K., Hou, S., Maeda, K., Minami, M. and Yanou, A., Analyses on on-line evolutionary optimization performance for pose tracking while eye-vergence visual servoing, Proceedings of 2013 IEEE International Conference on Mechatronics and Automation (ICMA) (2013), pp.698-703. 5ʣೲཅ, େଠ, ੴࢁ৽ଠ, ݟ࿘ޢ, ਫதࣗಈॆిΛࢦ͠ ͨϏδϡΞϧαʔϘܕਫதϩϘοτͷࣗಈቕ߹੍ޚ, ຊػց ֶձจू C ฤ (2015, ߘத) ʦஶ ऀ հʧ ೲ ཅʢਖ਼ձһʣ 1996 Ԭࢁେֶֶ෦ใֶՊଔۀɽ1998 ԬࢁେֶେֶӃֶڀݚՊम࢜՝ఔमྃɽ2001 ಉେֶେֶӃࣗવՊֶڀݚՊത࢜՝ఔमྃɽಉ େֶେֶӃڀݚੜΛͯܦɼ2002 ۙـେֶֶ෦ ॿखɽ2004 ಉߨࢣΛͯ ܦɼ2009 Ԭࢁେֶେ ֶӃࣗવՊֶڀݚՊॿڭɽࢸʹࡏݱΔɽ༧ଌ੍ޚ ʹؔ͢Δࣄैʹڀݚɽത࢜ʢֶʣ ɽܭଌࣗಈ੍ޚ ֶձɼγεςϜ੍ޚใֶձɼຊػցֶձͷ ձһɽ ܭଌࣗಈ੍ֶޚձจू ୈ xx רୈ xx ߸ 20xx xx ݄ Fig. 10 Tracking performance with additional disturbance made by air bubbles against dual-eye image recognition in the case that (A) 3D marker is moving on z-axis with amplitude 280 [mm] and period 20 [s] after 20 [s] and (B) disturbance is added after 10 [s]: (a)fitness value, (b)position of underwater vehicle in z-axis direction, (c)tracking error in z-axis direction, and (d)thrust in z-axis direction 7 Fig. 12 Tracking performance with additional disturbance made by air bubbles against dual-eye image recognition in the case that (A) 3D marker is moving on z-axis with amplitude 280 [mm] and period 10 [s] after 20 [s] and (B) disturbance is added after 10 [s]: (a)fitness value, (b)position of underwater vehicle in z-axis direction, (c)tracking error in z-axis direction, and (d)thrust in z-axis direction ੴ ࢁ ৽ଠʢਖ਼ձһʣ 1981 3 ݄౦ۀژେֶେֶӃ૯߹ཧֶڀݚ ՊΤωϧΪʔՊֶઐ߈म࢜՝ఔमྃ, 1990 6 ݄ ʹ౦ۀژେֶΑΓֶҐΛऔಘ, (ֶത࢜). 1981 4 ݄ຊॴڀݚྗࢠݪ౦ւߴॴڀݚԹֶ෦ೖ ॴ, 1993 ʙ1995 ϢʔϦοώॴڀݚʢυΠπʣ ʹͯࢠݪ༻৽ૉࡐ։ൃʹैࣄ, ΦʔΫδοϦࠃཱ ॴڀݚʢถʣʹͯ, ߴԹΨε༻ࠇԖࡐྉͷॏরࣹ ࣄैʹڀݚ. ຊֶྗࢠݪձϑΣʔϩʔ, 2010 4 ݄ʙ౦େࢠݪֶॴڀݚඇৗࢣߨۈ, 2012 4 ݄ʙࡏݱҪେֶԕ֎ྖҬڀݚηϯλʔ٬ һڭत, 2013 2 ݄ʙ2015 3 ݄Ԭࢁେֶࣗવ ՊֶڀݚՊ࢈ۀֶઐ߈ೳػցγεςϜֶ ٬һڭत. ݟ࿘ Fig. 11 Tracking performance with additional disturbance made by air bubbles against dual-eye image recognition in the case that (A) 3D marker is moving on z-axis with amplitude 280 [mm] and period 15 [s] after 20 [s] and (B) disturbance is added after 10 [s]: (a)fitness value, (b)position of underwater vehicle in z-axis direction, (c)tracking error in z-axis direction, and (d)thrust in z-axis direction ޢʢਖ਼ձһʣ 1979 େࡕཱେֶֶ෦ߤֶۭՊଔۀɼ1981 େࡕཱେֶֶڀݚՊߤֶۭઐ߈म࢜՝ఔ मྃɽ1993 ۚେֶେֶӃࣗવՊֶڀݚՊത ࢜՝ఔमྃɽത࢜ʢֶʣɽ1994 Ҫେֶֶ ෦ػցֶՊॿڭतɼ2002 ಉֶ෦ೳγεςϜ ֶՊڭतɼ2010 ԬࢁେֶେֶӃࣗવՊֶڀݚ ՊڭतɼࢸʹࡏݱΔɽϩϘοτͷྗֶɼ߆ଋӡಈɼ ྗ੍ޚɼҠಈϚχϐϡϨʔλͷ੍ޚɼը૾ೝࣝɼϏ δϡΞϧαʔϘΠϯάͷࣄैʹڀݚɽຊػց ֶ ձɼܭଌࣗಈ੍ֶޚձɼIEEE ͳͲͷձһɽ দ ོ ʢਖ਼ձһʣ ถ ݈ ଠ 2014 Ԭࢁେֶֶ෦γεςϜֶՊଔۀɽಉ ɼԬࢁେֶେֶӃࣗવՊֶڀݚՊػցγεςϜ ֶઐ߈ೖֶɼࢸʹࡏݱΔɽ 2004 9 ໊݄ݹେֶେֶӃֶڀݚՊϚΠ ΫϩγεςϜֶઐ߈ത࢜՝ఔظޙ՝ఔຬظୀֶɼ 2004 10 ໊݄ݹେֶେֶӃֶڀݚՊॿखɽ 2005 3 ݄ʹֶҐΛऔಘɼത࢜ (ֶ)ɽҎ߱ 2006 4 ݄ΑΓࢁཱݝେֶֶ෦ॿखɼॿڭɼߨࢣɽ 2011 10 ݄ΑΓԬࢁେֶେֶӃࣗવՊֶڀݚՊ ߨࢣʹணɽϩϘοτʹΑΔॊೈମͷϚχϐϡ Ϩʔγϣϯɼཱ࡞ۀͷࣗಈԽʹؔ͢Δࣄैʹڀݚɽ
© Copyright 2025 ExpyDoc
