%PDF-1.3 1 0 obj << /Kids [ 3 0 R 4 0 R 5 0 R 6 0 R 7 0 R 8 0 R 9 0 R 10 0 R 11 0 R 12 0 R ] /Type /Pages /Count 10 >> endobj 2 0 obj << /Title (Embodied Question Answering) /Producer (PyPDF2) /Author (Abhishek Das\054 Samyak Datta\054 Georgia Gkioxari\054 Stefan Lee\054 Devi Parikh\054 Dhruv Batra) /Subject (2018 IEEE Conference on Computer Vision and Pattern Recognition) >> endobj 3 0 obj << /Parent 1 0 R /Rotate 0 /Contents 14 0 R /Resources << /XObject << /x8 15 0 R /x6 18 0 R /x12 21 0 R /R34 24 0 R /x10 26 0 R /R33 25 0 R >> /ExtGState << /s9 29 0 R /R13 32 0 R /s11 33 0 R /a0 << /CA 1 /ca 1 >> /s5 36 0 R /s7 39 0 R >> /Font << /F2 42 0 R /R28 43 0 R /F1 47 0 R /R14 48 0 R /R16 51 0 R /R18 55 0 R /R20 58 0 R /R22 62 0 R /R24 66 0 R /R30 69 0 R /R26 72 0 R >> /ProcSet [ /Text /ImageC /ImageB /PDF /ImageI ] >> /Group 75 0 R /MediaBox [ 0 0 612 792 ] /Annots [ 76 0 R 77 0 R 78 0 R 79 0 R ] /Type /Page >> endobj 4 0 obj << /Parent 1 0 R /Rotate 0 /Type /Page /Contents 80 0 R /Resources << /ExtGState << /R13 32 0 R >> /Font << /F2 81 0 R /R28 43 0 R /F1 82 0 R /R14 48 0 R /R16 51 0 R /R26 72 0 R >> /ProcSet [ /ImageC /Text /PDF /ImageI /ImageB ] >> /MediaBox [ 0 0 612 792 ] /Annots [ 83 0 R 84 0 R 85 0 R 86 0 R 87 0 R 88 0 R 89 0 R 90 0 R 91 0 R 92 0 R 93 0 R ] >> endobj 5 0 obj << /Parent 1 0 R /Rotate 0 /Type /Page /Contents 94 0 R /Resources << /ColorSpace << /R94 95 0 R >> /ExtGState << /R13 32 0 R >> /Font << /F2 97 0 R /R99 98 0 R /F1 102 0 R /R14 48 0 R /R16 51 0 R /R28 43 0 R /R101 103 0 R /R95 107 0 R /R97 111 0 R >> /ProcSet [ /ImageC /Text /PDF /ImageI /ImageB ] >> /MediaBox [ 0 0 612 792 ] /Annots [ 115 0 R 116 0 R 117 0 R 118 0 R 119 0 R 120 0 R 121 0 R 122 0 R 123 0 R 124 0 R 125 0 R 126 0 R 127 0 R 128 0 R 129 0 R 130 0 R 131 0 R 132 0 R 133 0 R 134 0 R 135 0 R 136 0 R 137 0 R 138 0 R 139 0 R 140 0 R 141 0 R 142 0 R 143 0 R 144 0 R 145 0 R 146 0 R 147 0 R 148 0 R ] >> endobj 6 0 obj << /Parent 1 0 R /Rotate 0 /Type /Page /Contents 149 0 R /Resources << /XObject << /R123 150 0 R /R122 151 0 R >> /ExtGState << /R124 152 0 R /R121 153 0 R /R120 154 0 R >> /Font << /F2 155 0 R /R131 156 0 R /R133 159 0 R /F1 164 0 R /R127 165 0 R /R99 98 0 R /R125 169 0 R /R28 43 0 R /R14 48 0 R /R135 173 0 R /R16 51 0 R /R137 177 0 R /R129 180 0 R /R26 72 0 R >> /ProcSet [ /Text /ImageC /ImageB /PDF /ImageI ] /ColorSpace << /R94 95 0 R >> >> /MediaBox [ 0 0 612 792 ] /Annots [ 183 0 R 184 0 R 185 0 R 186 0 R 187 0 R 188 0 R 189 0 R 190 0 R 191 0 R 192 0 R 193 0 R 194 0 R ] >> endobj 7 0 obj << /Parent 1 0 R /Rotate 0 /Type /Page /Contents 195 0 R /Resources << /ColorSpace << /R94 95 0 R >> /ExtGState << /R13 32 0 R /R124 152 0 R /R121 153 0 R >> /Font << /R156 196 0 R /R154 200 0 R /R152 204 0 R /R99 98 0 R /F1 208 0 R /F2 209 0 R /R28 43 0 R /R133 159 0 R /R14 48 0 R /R16 51 0 R /R137 177 0 R /R129 180 0 R /R26 72 0 R >> /ProcSet [ /ImageC /Text /PDF /ImageI /ImageB ] >> /MediaBox [ 0 0 612 792 ] /Annots [ 210 0 R 211 0 R 212 0 R 213 0 R 214 0 R ] >> endobj 8 0 obj << /Parent 1 0 R /Rotate 0 /Type /Page /Contents 215 0 R /Resources << /XObject << /R181 216 0 R /R180 218 0 R /R183 219 0 R /R182 220 0 R /R185 221 0 R /R184 222 0 R /R187 223 0 R /R186 224 0 R /R178 225 0 R /R179 226 0 R /R174 227 0 R /R175 228 0 R /R176 217 0 R /R177 229 0 R /R170 230 0 R /R173 231 0 R >> /ExtGState << /R121 153 0 R /R120 154 0 R >> /Font << /R190 232 0 R /R152 204 0 R /R131 156 0 R /R101 103 0 R /R188 235 0 R /R133 159 0 R /R168 238 0 R /R171 242 0 R /F1 246 0 R /F2 247 0 R /R99 98 0 R /R28 43 0 R /R14 48 0 R /R16 51 0 R /R129 180 0 R >> /ProcSet [ /Text /ImageC /ImageB /PDF /ImageI ] /ColorSpace << /R94 95 0 R >> >> /MediaBox [ 0 0 612 792 ] /Annots [ 248 0 R 249 0 R 250 0 R 251 0 R ] >> endobj 9 0 obj << /Parent 1 0 R /Rotate 0 /Type /Page /Contents 252 0 R /Resources << /XObject << /R202 253 0 R /R201 254 0 R >> /ExtGState << /R13 32 0 R >> /Font << /R203 255 0 R /R207 258 0 R /R205 261 0 R /R101 103 0 R /R99 98 0 R /R131 156 0 R /F1 265 0 R /F2 266 0 R /R28 43 0 R /R133 159 0 R /R14 48 0 R /R16 51 0 R /R129 180 0 R /R26 72 0 R >> /ProcSet [ /Text /ImageC /ImageB /PDF /ImageI ] >> /MediaBox [ 0 0 612 792 ] /Annots [ 267 0 R 268 0 R 269 0 R ] >> endobj 10 0 obj << /Parent 1 0 R /Rotate 0 /Type /Page /Contents 270 0 R /Resources << /ExtGState << /R13 32 0 R >> /Font << /R203 255 0 R /R207 258 0 R /R205 261 0 R /R188 235 0 R /R222 271 0 R /R224 275 0 R /F2 278 0 R /R99 98 0 R /R133 159 0 R /R129 180 0 R /R152 204 0 R /R131 156 0 R /R101 103 0 R /R28 43 0 R /F1 279 0 R /R14 48 0 R /R16 51 0 R /R26 72 0 R >> /ProcSet [ /ImageC /Text /PDF /ImageI /ImageB ] >> /MediaBox [ 0 0 612 792 ] /Annots [ 280 0 R 281 0 R 282 0 R 283 0 R 284 0 R 285 0 R 286 0 R ] >> endobj 11 0 obj << /Parent 1 0 R /Rotate 0 /Type /Page /Contents 287 0 R /Resources << /ExtGState << /R13 32 0 R >> /Font << /F2 288 0 R /R28 43 0 R /F1 289 0 R /R14 48 0 R /R16 51 0 R >> /ProcSet [ /ImageC /Text /PDF /ImageI /ImageB ] >> /MediaBox [ 0 0 612 792 ] /Annots [ 290 0 R 291 0 R 292 0 R 293 0 R 294 0 R 295 0 R 296 0 R 297 0 R 298 0 R 299 0 R 300 0 R 301 0 R 302 0 R 303 0 R 304 0 R 305 0 R 306 0 R 307 0 R 308 0 R 309 0 R 310 0 R 311 0 R 312 0 R 313 0 R 314 0 R 315 0 R 316 0 R 317 0 R 318 0 R 319 0 R 320 0 R 321 0 R 322 0 R 323 0 R 324 0 R 325 0 R 326 0 R 327 0 R 328 0 R 329 0 R 330 0 R 331 0 R 332 0 R 333 0 R 334 0 R 335 0 R 336 0 R 337 0 R 338 0 R ] >> endobj 12 0 obj << /Parent 1 0 R /Rotate 0 /Type /Page /Contents 339 0 R /Resources << /ExtGState << /R13 32 0 R >> /Font << /F2 340 0 R /R28 43 0 R /F1 341 0 R /R16 51 0 R /R26 72 0 R >> /ProcSet [ /ImageC /Text /PDF /ImageI /ImageB ] >> /MediaBox [ 0 0 612 792 ] /Annots [ 342 0 R 343 0 R 344 0 R 345 0 R 346 0 R 347 0 R 348 0 R 349 0 R 350 0 R ] >> endobj 13 0 obj << /Type /Catalog /Pages 1 0 R >> endobj 14 0 obj << /Length 16118 >> stream q q q 0.1 0 0 0.1 0 0 cm /R13 gs 0 g q 10 0 0 10 0 0 cm BT /R14 14.3462 Tf 1 0 0 1 201.96 675.426 Tm [ (Embodied) -250.002 (Question) -249.996 (Answering) ] TJ /R16 11.9552 Tf -145.87 -37.5789 Td [ (Abhishek) -249.984 (Das) ] TJ /R18 7.9701 Tf 67.4023 4.33789 Td [ (1) -0.29866 ] TJ /R20 5.9776 Tf 4.23398 2.81289 Td <0e> Tj /R16 11.9552 Tf 4.64883 -7.15078 Td [ (\054) -833.015 (Samyak) -250.012 (Datta) ] TJ /R18 7.9701 Tf 80.3539 4.33789 Td [ (1) -0.30019 ] TJ /R16 11.9552 Tf 4.7332 -4.33789 Td [ (\054) -833.015 (Geor) 18.0034 (gia) -250 (Gkioxari) ] TJ /R18 7.9701 Tf 96.7328 4.33789 Td [ (2) -0.30019 ] TJ /R16 11.9552 Tf 4.73203 -4.33789 Td [ (\054) -833.015 (Stef) 10.0044 (an) -250.008 (Lee) ] TJ /R18 7.9701 Tf 64.2879 4.33789 Td [ (1) -0.30019 ] TJ /R16 11.9552 Tf 4.73203 -4.33789 Td [ (\054) -833.015 (De) 24.9875 (vi) -249.993 (P) 15.0137 (arikh) ] TJ /R18 7.9701 Tf 69.918 4.33789 Td [ (2) -0.30019 ] TJ /R22 7.9701 Tf 4.23438 0 Td [ (\073) -0.09802 ] TJ /R18 7.9701 Tf 2.35273 0 Td [ (1) -0.30019 ] TJ /R16 11.9552 Tf 4.73203 -4.33789 Td [ (\054) -833.013 (Dhruv) -250.002 (Batra) ] TJ /R18 7.9701 Tf 72.3809 4.33789 Td [ (2) -0.30019 ] TJ /R22 7.9701 Tf 4.23438 0 Td [ (\073) -0.09802 ] TJ /R18 7.9701 Tf 2.35195 0 Td [ (1) -0.30019 ] TJ -380.195 -16.657 Td [ (1) -0.29866 ] TJ /R16 11.9552 Tf 4.73203 -4.33906 Td [ (Geor) 18.0044 (gia) -250 (Institute) -250.016 (of) -250.014 (T) 70.0164 (echnology) 65.0132 (\054) ] TJ /R18 7.9701 Tf 157.879 4.33906 Td [ (2) -0.30019 ] TJ /R16 11.9552 Tf 4.73164 -4.33906 Td [ (F) 15.0158 (acebook) -249.993 (AI) -250.002 (Research) ] TJ /R24 5.9776 Tf -231.925 -9.85898 Td [ (1) -0.09802 ] TJ /R26 8.9664 Tf 4.15117 -3.80898 Td [ (\173abhshkdz\054) -599.987 (samyak\054) -599.994 (steflee\175\100gatech\056edu) ] TJ /R24 5.9776 Tf 219.378 3.80898 Td [ (2) -0.09802 ] TJ /R26 8.9664 Tf 4.15078 -3.80898 Td [ (\173gkioxari\054) -599.988 (parikh\054) -599.994 (dbatra\175\100fb\056com) ] TJ ET Q 0 1 0 0 k q 10 0 0 10 0 0 cm BT /R26 9.9626 Tf 1 0 0 1 264.138 593.852 Tm (embodiedqa\056org) Tj ET Q 0 g q 10 0 0 10 0 0 cm BT /R14 11.9552 Tf 1 0 0 1 145.995 552.89 Tm (Abstract) Tj /R28 9.9626 Tf -95.8828 -26.6602 Td [ (W) 91.9865 (e) -238.004 (pr) 36.9852 (esent) -239.009 (a) -237.995 (ne) 15.0183 (w) -237.986 (AI) -239.018 (task) -237.986 (\226) ] TJ /R30 9.9626 Tf 108.703 0 Td [ (Embodied) -237.991 (Question) -239.007 (Answ) 10.0057 (ering) ] TJ /R28 9.9626 Tf -108.703 -11.716 Td [ (\050EmbodiedQA\051) -333.993 (\226) -334.998 (wher) 36.9938 (e) -333.988 (an) -334.993 (a) 10.0032 (g) 10.0032 (ent) -334.015 (is) -334.009 (spawned) -335.008 (at) -334.01 (a) -334.998 (r) 14.9828 (andom) ] TJ 11.716 TL T* [ (location) -281.994 (i) 0.98758 (n) -281.987 (a) -281.987 (3D) -281.005 (en) 39.9946 (vir) 44.9815 (onment) -282.011 (and) -280.997 (ask) 10.0057 (ed) -281.992 (a) -281.007 (question) -281.995 (\050) ] TJ /R16 9.9626 Tf 211.354 0 Td (\140What) Tj -211.354 -11.716 Td [ (color) -189.998 (is) -189.994 (the) -191.019 (car\077\047) ] TJ /R28 9.9626 Tf 64.884 0 Td [ (\051\056) -289.992 (In) -190.006 (or) 36.9852 (der) -190.98 (to) -189.995 (answer) 110.993 (\054) -202.003 (the) -190 (a) 10.0032 (g) 10.0032 (ent) -190 (must) -191.005 <0272> 10.0106 (st) -189.995 (in\055) ] TJ -64.884 -11.716 Td [ (tellig) 9.98853 (ently) -263.005 (navigate) -262.009 (to) -263.003 (e) 19.9918 (xplor) 36.9926 (e) -263.02 (the) -261.988 (en) 39.9946 (vir) 44.9827 (onment\054) -266.012 (gather) -262.99 (nec\055) ] TJ T* [ (essary) -296.995 (visual) -298.002 (information) -296.986 (thr) 44.9913 (ough) -296.99 <0272> 10.0094 (st\055per) 10.0155 (son) -298.012 (\050e) 39.986 (gocentric\051) ] TJ T* [ (vision\054) -250.008 (and) -249.982 (then) -249.985 (answer) -250.006 (the) -249.99 (question) -250.001 (\050) ] TJ /R16 9.9626 Tf 151.371 0 Td (\140orange\047) Tj /R28 9.9626 Tf 33.743 0 Td (\051\056) Tj -185.114 -14.7047 Td [ (EmbodiedQA) -327.007 (r) 37.0196 (equir) 36.9926 (es) -327.983 (a) -327 (r) 14.984 (ang) 10.013 (e) -327.009 (of) -328.011 (AI) -327.003 (skills) -327.014 (\226) -327 (langua) 9.99098 (g) 10.0032 (e) -327.989 (un\055) ] TJ T* [ (der) 10.0051 (standing) 10.0045 (\054) -525.008 (visual) -468.994 (r) 37.0196 (eco) 9.99466 (gnition\054) -525.006 (active) -469.993 (per) 36.9816 (ception\054) -524.988 (goal\055) ] TJ T* [ (driven) -679.009 (navigation\054) -786.015 (commonsense) -679 (r) 37.0196 (easoning) 9.99343 (\054) -785.984 (long\055term) ] TJ T* [ (memory) 55.0184 (\054) -214.981 (and) -207.009 (gr) 44.9839 (ounding) -206.988 (langua) 9.99098 (g) 10.0032 (e) -206.01 (into) -207 (actions\056) -295.014 (In) -206.983 (this) -207 (work\054) ] TJ T* [ (we) -249.008 (de) 15.0177 (velop) -249.009 (a) -249.012 (dataset) -248.006 (of) -249.005 (questions) -249.015 (and) -249.003 (answer) 9.99098 (s) -249.012 (in) -249.005 (House3D) ] TJ T* [ (en) 39.9939 (vir) 44.9821 (onments) -419.981 (\133) ] TJ ET Q 0.75 0 0.25 rg q 10 0 0 10 0 0 cm BT /R28 9.9626 Tf 1 0 0 1 111.004 382.649 Tm (1) Tj ET Q 0 g q 10 0 0 10 0 0 cm BT /R28 9.9626 Tf 1 0 0 1 115.985 382.649 Tm [ (\135\054) -461.993 (e) 15.0122 (valuation) -420.006 (metrics\054) -462.999 (and) -419.994 (a) -420.003 (hier) 14.9926 (ar) 36.9865 (c) 15.0122 (hical) ] TJ -65.8727 -11.716 Td [ (model) -249.982 (tr) 14.9914 (ained) -250.018 (with) -250.004 (imitation) -249.983 (and) -249.983 (r) 37.0196 (einfor) 37.0012 (cement) -249.995 (learning) 15.0147 (\056) ] TJ /R14 11.9552 Tf 37.6281 TL T* [ (1\056) -249.99 (Intr) 18.0146 (oduction) ] TJ /R16 8.9664 Tf 35.4359 -23.159 Td [ (The) -263.987 (embodiment) -264.009 (h) 5.01002 (ypothesis) -264 (is) -263.993 (the) -264.017 (idea) -263.993 (t) 1.01017 (hat) -264.017 (intelligence) ] TJ 10.7398 TL T* [ (emer) 17.9925 (ges) -309.983 (in) -311.005 (the) -310.021 (interaction) -310.003 (of) -309.983 (an) -310.989 (agent) -309.998 (with) -310.003 (an) -309.987 (en) 40.0176 (viron\055) ] TJ 10.7391 TL T* [ (ment) -249.98 (and) -249.997 (as) -249.993 (a) -249.997 (result) -249.989 (of) -249.993 (sensorimotor) -249.992 (acti) 24.9875 (vity) 65.0159 (\056) ] TJ ET Q 3.98 w 0 G 855.48 2801.44 m 2863.62 2801.44 l S q 10 0 0 10 0 0 cm BT /R28 8.9664 Tf 1 0 0 1 208.902 270.851 Tm [ (Smith) -250.014 (and) -249.978 (Gasser) -249.995 (\133) ] TJ ET Q 0.75 0 0.25 rg q 10 0 0 10 0 0 cm BT /R28 8.9664 Tf 1 0 0 1 278.391 270.851 Tm (2) Tj ET Q 0 g q 10 0 0 10 0 0 cm BT /R28 8.9664 Tf 1 0 0 1 282.875 270.851 Tm (\135) Tj /R16 9.9626 Tf -232.762 -17.7531 Td [ (Our) -390.009 (long\055term) -391.002 (goal) -390.001 (is) -390 (to) -390.981 (b) 20.0016 (uild) -389.989 (intelligent) -390.004 (agents) -389.999 (that) -391.018 (can) ] TJ /R28 9.9626 Tf 11.716 TL T* [ (per) 36.981 (ceive) ] TJ /R16 9.9626 Tf 38.1879 0 Td [ (their) -483.005 (en) 39.9946 (vironment) -483.017 (\050through) -484.011 (vision\054) -541.01 (audition\054) -541.988 (or) ] TJ -38.1879 -11.716 Td [ (other) -334.993 (sensors\051\054) ] TJ /R28 9.9626 Tf 62.4918 0 Td (communicate) Tj /R16 9.9626 Tf 56.4562 0 Td (\050) Tj /R28 9.9626 Tf 3.3168 0 Td (i\056e) Tj /R16 9.9626 Tf 9.68398 0 Td [ (\056\054) -356.014 (hold) -334.981 (a) -335.008 (natur) 0.99003 (al) -335 (language) ] TJ -131.949 -11.716 Td [ (dialog) -314 (grounded) -312.985 (in) -314.014 (the) -313 (en) 39.9946 (vironment\051\054) -329.985 (and) ] TJ /R28 9.9626 Tf 171.517 0 Td (act) Tj /R16 9.9626 Tf 15.477 0 Td (\050) Tj /R28 9.9626 Tf 3.31797 0 Td [ (e) 15.0122 (\056g) ] TJ /R16 9.9626 Tf 11.7461 0 Td [ (\056) -313.004 (aid) -314.019 (hu\055) ] TJ -202.058 -11.716 Td [ (mans) -346.982 (by) -346.991 (e) 15.0128 (x) 14.9828 (ecuting) -347.019 (API) -346.007 (calls) -346.989 (or) -347 (commands) -347.003 (in) -346.989 (a) -347.006 (virtual) -347.016 (or) ] TJ T* [ (embodied) -303.017 (en) 39.9939 (vironment\051\056) -468.002 (In) -303.007 (addition) -301.991 (to) -302.996 (being) -302.991 (a) -303.013 (fundamen\055) ] TJ T* [ (tal) -241.989 (scie) 0.98452 <6e74690263> -242.009 (goal) -241.987 (in) -241.006 <61727469026369616c> -242.011 (intelligence) -240.984 (\050AI\051\054) -242.004 (e) 25.0105 (v) 14.9828 (en) -241.019 (a) -242.004 (small) ] TJ T* [ (adv) 24.9817 (ance) -194.015 (to) 24.9885 (w) 10.0014 (ards) -194.983 (such) -193.995 (intelligent) -193.997 (systems) -194.007 (can) ] TJ /R28 9.9626 Tf 179.244 0 Td (fundamentally) Tj -179.244 -11.716 Td [ (c) 15.0128 (hang) 10.0185 (e) -327.989 (our) -328.008 (lives) ] TJ /R16 9.9626 Tf 70.4309 0 Td [ (\226) -328.019 (from) -328.013 (assisti) 25.0118 (v) 14.9828 (e) -327.989 (dialog) -329.019 (agents) -328.009 (for) -327.987 (the) -328.016 (vi\055) ] TJ -70.4309 -11.716 Td [ (sually) -359.992 (impaired\054) -387.987 (to) -360.986 (natural\055language) -359.995 (interaction) -359.991 (with) -361.016 (self\055) ] TJ T* [ (dri) 24.9854 (ving) -250.015 (cars\054) -250 (in\055home) -250.01 (robots\054) -250.01 (and) -249.991 (personal) -249.98 (assistants\056) ] TJ 14.7051 TL T* [ (As) -193.017 (a) -194.011 (step) -193.013 (to) 24.9885 (w) 10.0014 (ards) -194.003 (goal) 0.99248 (\055dri) 24.9824 (v) 14.9828 (en) -194.006 (agents) -193.012 (that) -193.992 (can) -192.99 (percei) 25.0105 (v) 14.9828 (e\054) -204.988 (com\055) ] TJ 11.716 TL T* [ (municate\054) -407.004 (and) -374.99 (e) 15.0122 (x) 14.9828 (ecute) -376.017 (actions\054) -407.006 (we) -374.997 (present) -374.984 (a) -375.981 (ne) 25.0154 (w) -374.994 (AI) -376.016 (task) ] TJ (\226) ' /R30 9.9626 Tf 8.21094 0 Td [ (Embodied) -324.015 (Question) -324.011 (Answ) 10.0057 (ering) ] TJ /R16 9.9626 Tf 132.486 0 Td [ (\050EmbodiedQA\051) -323.985 (\226) -324.019 (along) ] TJ ET Q 501.121 904.148 m 1446.11 904.148 l S q 10 0 0 10 0 0 cm BT /R20 5.9776 Tf 1 0 0 1 60.3078 83.8129 Tm <0e> Tj /R16 7.9701 Tf 4.15 -2.81289 Td [ (W) 79.9796 (ork) -250.016 (partially) -250.02 (done) -249.991 (during) -249.987 (an) -250.013 (internship) -250.016 (at) -249.993 (F) 15.0128 (acebook) -249.991 (AI) -250.025 (Research\056) ] TJ ET Q q 2362.52 0 0 1153.7 3088.62 4474.84 cm /R34 Do Q q 10 0 0 10 0 0 cm BT /R16 8.9664 Tf 1 0 0 1 308.862 424.869 Tm [ (Figure) -311.019 (1\072) -430.985 (Embodied) -310.02 (Question) -311.019 (Answering) -310.006 (\226) -311.014 (EmbodiedQA\226) -309.995 (tasks) ] TJ 10.959 TL T* [ (agents) -348.984 (with) -348.994 (na) 20.021 (vig) 5.01819 (ating) -348 (rich) -348.984 (3D) -349.013 (en) 40.0148 (vironments) -349.008 (in) -348.992 (order) -349.008 (to) -347.99 (answer) ] TJ T* [ (questions\056) -537.004 (These) -324.993 (agents) -325.981 (must) -325.978 (jointly) -325.009 (learn) -326 (language) -325.003 (understand\055) ] TJ T* [ (ing\054) -249.992 (visual) -250.019 (reasoning\054) -249.997 (and) -249.997 (goal\055dri) 25.0229 (v) 14.9865 (en) -250 (na) 20.0183 (vig) 5.02092 (ation) -249.978 (to) -250.011 (succeed\056) ] TJ /R16 9.9626 Tf 34.1723 TL T* [ (with) -270.011 (a) -271.018 (dataset) -269.982 (of) -269.992 (questions) -271.009 (in) -269.984 (virtual) -270.009 (en) 39.9933 (vironments\054) -275.983 (e) 25.0105 (v) 24.9811 (alua\055) ] TJ 11.716 TL T* [ (tion) -199.981 (metrics\054) -209.002 (and) -200 (a) -200.011 (deep) -200.003 (reinforcement) -199.011 (learning) -199.986 (\050RL\051) -199.996 (model\056) ] TJ 14.7051 TL T* [ (Concretely) 64.9965 (\054) -313.984 (the) -301.001 (EmbodiedQA) -301.004 (task) -302.013 (is) -300.994 (illustrated) -301.009 (in) -300.999 (Fig\056) ] TJ ET Q 0.75 0 0.25 rg q 10 0 0 10 0 0 cm BT /R16 9.9626 Tf 1 0 0 1 532.144 331.399 Tm (1) Tj ET Q 0 g q 10 0 0 10 0 0 cm BT /R16 9.9626 Tf 1 0 0 1 540.134 331.399 Tm (\226) Tj -231.272 -11.716 Td [ (an) -226.982 (agent) -227.994 (is) -227.009 (spa) 14.9852 (wned) -227.014 (at) -227.018 (a) -228.006 (random) -227.001 (location) -226.992 (in) -227.011 (an) -228.001 (en) 39.9933 (vironment) ] TJ 11.7156 TL T* [ (\050a) -226.009 (house) -227.004 (or) -226 (b) 20.0016 (uilding\051) -226.001 (and) -225.997 (ask) 10.0032 (ed) -226.982 (a) -226.006 (question) -226.006 (\050) ] TJ /R28 9.9626 Tf 172.125 0 Td [ (e) 15.0122 (\056g) ] TJ /R16 9.9626 Tf 11.7461 0 Td (\056) Tj /R28 9.9626 Tf 4.74492 0 Td [ (\140What) -225.987 (color) ] TJ -188.616 -11.716 Td [ (is) -298.997 (the) -299.002 (car\077\047) ] TJ /R16 9.9626 Tf 46.3508 0 Td [ (\051\056) -456.002 (The) -299.004 (agent) -299.002 (percei) 25.0105 (v) 14.9828 (es) -299.009 (its) -298.009 (en) 39.9933 (vironment) -299.009 (through) ] TJ -46.3508 -11.716 Td [ <027273742d706572736f6e> -234.005 (e) 15.0122 (gocentric) -233.988 (vision) -232.986 (and) -233.995 (can) -234.005 (perform) -233.99 (a) -234.005 (fe) 25.0056 (w) -233.995 (atomic) ] TJ T* [ (actions) -332.981 (\050mo) 14.9926 (v) 14.9828 (e\055forw) 9.99098 (ard\054) -354.019 (turn\054) -354.005 (strafe\054) ] TJ /R28 9.9626 Tf 147.679 0 Td (etc) Tj /R16 9.9626 Tf 11.616 0 Td [ (\056\051\056) -559.985 (The) -332.998 (goal) -332.989 (of) -333.003 (the) ] TJ -159.295 -11.716 Td [ (agent) -215.996 (is) -217.01 (to) -215.991 (intelligently) -215.981 (na) 19.9967 (vig) 5 (ate) -216.983 (the) -215.993 (en) 39.9933 (vironment) -216.003 (and) -217.018 (g) 4.98446 (ather) ] TJ T* [ (visual) -250.01 (information) -249.993 (necessary) -249.997 (for) -249.997 (answering) -250.017 (the) -249.988 (question\056) ] TJ 14.7051 TL T* [ (EmbodiedQA) -370.012 (is) -370.004 (a) -369.982 (challenging) -369.982 (task) -370.002 (that) -370.002 (subsumes) -370.017 (se) 25.0154 (v) 14.9828 (eral) ] TJ 11.716 TL T* [ (fundamental) -236.01 (problems) -236.987 (as) -235.997 (sub\055tasks\056) -305.013 (Clearly) 64.9892 (\054) -239.014 (the) -235.99 (agent) -237.009 (must) ] TJ T* [ (understand) -411.983 (language) -413.014 (\050) ] TJ /R28 9.9626 Tf 91.216 0 Td [ (what) -412.019 (is) -413.014 (the) -412 (question) -412.995 (asking\077) ] TJ /R16 9.9626 Tf 119.388 0 Td [ (\051) -797.009 (and) ] TJ -210.604 -11.716 Td [ (vision) -250.002 (\050) ] TJ /R28 9.9626 Tf 30.166 0 Td [ (what) -250.007 (does) -249.985 (a) -249.993 (\140car\047) -249.997 (look) -249.983 (lik) 10.0081 (e\077) ] TJ /R16 9.9626 Tf 111.412 0 Td [ (\051\054) -249.997 (b) 20.0016 (ut) -249.988 (it) -249.983 (must) -250.007 (also) -249.983 (learn\072) ] TJ /R14 9.9626 Tf -138.035 -16.4582 Td [ (Acti) 10 (v) 10.0032 (e) -234.005 (P) 20.0016 (er) 17.9872 (ception) ] TJ /R16 9.9626 Tf 74.793 0 Td [ (\072) -301.981 (The) -235.015 (agent) -233.99 (may) -233.985 (be) -235.02 (spa) 14.9828 (wned) -233.99 (an) 15.0171 (ywhere) ] TJ -74.793 -11.7156 Td [ (in) -195.994 (the) -197.019 (en) 39.9933 (vironment) -196.007 (and) -196.982 (may) -195.992 (not) -195.987 (immediately) -197.016 (\140see\047) -196.016 (the) -195.997 (pix\055) ] TJ T* [ (els) -340.012 (containing) -339.985 (the) -340.014 (answer) -339.982 (to) -340.012 (the) -340.012 (visual) -339.997 (question) -339.987 (\050) ] TJ /R28 9.9626 Tf 201.427 0 Td (i\056e) Tj /R16 9.9626 Tf 9.68398 0 Td [ (\056) -340.017 (the) ] TJ -211.111 -11.716 Td [ (car) -210.016 (may) -209.992 (not) -211.006 (be) -210.004 (visible\051\056) -295.997 (Thus\054) -218.998 (the) -209.994 (agent) ] TJ /R28 9.9626 Tf 157.697 0 Td (must) Tj /R16 9.9626 Tf 21.0922 0 Td [ (mo) 14.9975 (v) 14.9828 (e) -210.009 (to) -209.994 (suc\055) ] TJ -178.789 -11.716 Td [ (ceed) -305.996 (\226) -305.003 (controlling) -305.981 (the) -306.02 (pix) 14.9975 (els) -304.998 (that) -306.013 (it) -306.013 (percei) 25.0105 (v) 14.9828 (es\056) -477.014 (The) -305.003 (agent) ] TJ T* [ (must) -355.99 (learn) -356.019 (to) -356.007 (map) -356.007 (its) -356 (visual) -355.995 (input) -355.99 (to) -356.009 (the) -356.009 (corr) 1.01454 (ect) -355.98 (actions) ] TJ T* [ (based) -239.992 (on) -239.009 (its) -240.019 (perception) -239.982 (of) -240 (the) -239.009 (w) 10.0032 (orld\054) -241.984 (the) -239.989 (underlying) -238.98 (ph) 4.98936 (ys\055) ] TJ T* [ (ical) -249.993 (constraints\054) -249.997 (and) -249.993 (its) -250.017 (understanding) -249.993 (of) -249.997 (the) -249.988 (question\056) ] TJ ET Q Q Q q q 1 1 1 rg /a0 gs 48.406 786.422 515.188 -52.699 re f q /s5 gs /x6 Do Q q /s7 gs /x8 Do Q q /s9 gs /x10 Do Q q /s11 gs /x12 Do Q Q Q Q q 1 0 0 1 0 0 cm BT /F1 12 Tf 14.4 TL ET 1 1 1 rg n 270 32 72 14 re f* 0.5 0.5 0.5 rg BT /F2 9 Tf 10.8 TL ET BT 1 0 0 1 303.75 35 Tm (1) Tj T* ET Q endstream endobj 15 0 obj << /Filter /FlateDecode /Resources << /ExtGState << /a0 << /CA 1 /ca 1 >> >> /XObject << /x18 16 0 R >> >> /Length 28 /Group << /Type /Group /S /Transparency /CS /DeviceRGB /I true >> /BBox [ 78 746 96 765 ] /Type /XObject /Subtype /Form >> stream x+O4PH/VЯ0Pp 0 endstream endobj 16 0 obj << /Filter /FlateDecode /Resources 17 0 R /Length 107 /Type /XObject /BBox [ 78 746 96 765 ] /Subtype /Form >> stream xe AC̬wʠ =p,?]%+H-
Jc "82w8VSnGW;"
endstream
endobj
17 0 obj
<<
/ExtGState <<
/a0 <<
/CA 1
/ca 1
>>
>>
>>
endobj
18 0 obj
<<
/Filter /FlateDecode
/Resources <<
/ExtGState <<
/a0 <<
/CA 1
/ca 1
>>
>>
/XObject <<
/x15 19 0 R
>>
>>
/Length 28
/Group <<
/Type /Group
/S /Transparency
/CS /DeviceRGB
/I true
>>
/BBox [ 67 752 84 775 ]
/Type /XObject
/Subtype /Form
>>
stream
x+O4PH/VЯ04Up
0
endstream
endobj
19 0 obj
<<
/Filter /FlateDecode
/Resources 20 0 R
/Length 228
/Type /XObject
/BBox [ 67 752 84 775 ]
/Subtype /Form
>>
stream
xeQKn!s ?FPav6R٪TS.
b];15YyR
{7QL.\:Rv/x9l+L7h%1!}i/AI(kz"U&,YO![R hg{3}4/GyYF:!w}Gn+'xJcO9i뽼_-:`
endstream
endobj
20 0 obj
<<
/ExtGState <<
/a0 <<
/CA 1
/ca 1
>>
>>
>>
endobj
21 0 obj
<<
/Filter /FlateDecode
/Resources <<
/ExtGState <<
/a0 <<
/CA 1
/ca 1
>>
>>
/XObject <<
/x24 22 0 R
>>
>>
/Length 28
/Group <<
/Type /Group
/S /Transparency
/CS /DeviceRGB
/I true
>>
/BBox [ 132 751 480 772 ]
/Type /XObject
/Subtype /Form
>>
stream
x+O4PH/VЯ02Qp
0
endstream
endobj
22 0 obj
<<
/Filter /FlateDecode
/Resources 23 0 R
/Length 53223
/Type /XObject
/BBox [ 132 751 480 772 ]
/Subtype /Form
>>
stream
xtI:6%Q㨈?7rA= u%6 ?Y(WbWo{B>9
x`Znϳ|8{3?0x*z ǃ|,@:w>`c|*ϻⳅKO3`g
:_|}}><.6`Z{{3]#<_o"~:ͺgk7/Ұ@|K yp ]03ʷCmş8˽Y?>(3!Bwqs.Z8,~~=rMT̩y+/*w: uBZ_`ߵp`%M?ɝ1ɳw=vDۉy&xb4Q>d@ sg~lA