• Àüü
  • ÀüÀÚ/Àü±â
  • Åë½Å
  • ÄÄÇ»ÅÍ
´Ý±â

»çÀÌÆ®¸Ê

Loading..

Please wait....

±¹³» ³í¹®Áö

Ȩ Ȩ > ¿¬±¸¹®Çå > ±¹³» ³í¹®Áö > Çѱ¹Á¤º¸°úÇÐȸ ³í¹®Áö > Á¤º¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)

Á¤º¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)

Current Result Document : 3 / 91 ÀÌÀü°Ç ÀÌÀü°Ç   ´ÙÀ½°Ç ´ÙÀ½°Ç

ÇѱÛÁ¦¸ñ(Korean Title) °¡»ó ½ºÄµ ¸ÅĪÀ» ÀÌ¿ëÇÑ 3Â÷¿ø Á¡±º ÁöµµÀÇ Ç°Áú Æò°¡
¿µ¹®Á¦¸ñ(English Title) Quality Evaluation of 3D Point Maps using Virtual Scan Matching
ÀúÀÚ(Author) ¹éÇѳª   ±è°­Èñ   Hannah Baek   Kanghee Kim  
¿ø¹®¼ö·Ïó(Citation) VOL 27 NO. 02 PP. 0083 ~ 0089 (2021. 02)
Çѱ۳»¿ë
(Korean Abstract)
3Â÷¿ø Á¡±º Áöµµ´Â ÀÚÀ²ÁÖÇàÂ÷°¡ ½Ç½Ã°£À¸·Î ¼ö½ÅÇÏ´Â ¶óÀÌ´õ ½ºÄµ°ú ¸ÅĪ(matching)ÇÔÀ¸·Î½á Â÷·®ÀÇ À§Ä¡¸¦ ¼¾Æ¼¹ÌÅÍ ¼öÁØÀÇ Á¤¹Ðµµ·Î ÃßÁ¤Çϴµ¥ »ç¿ëµÈ´Ù. ±×·¯³ª, 3Â÷¿ø Á¡±º Áöµµ´Â ´Ù¾çÇÑ Á¶°Ç°ú »óȲ¿¡¼­ ¼ö½ÅµÈ Â÷·®ÀÇ ½Ç½Ã°£ ½ºÄµ¿¡ ´ëÇؼ­ Ç×»ó Á¤È®ÇÑ ÀÚÂ÷ À§Ä¡ ÃßÁ¤ °á°ú¸¦ º¸ÀåÇÏÁö´Â ¾Ê´Â´Ù. ¿Ö³ÄÇϸé, Â÷·®ÀÇ ½Ç½Ã°£ ½ºÄµÀº ¶óÀÌ´õ ¸ðµ¨, Â÷·®¿¡ ÀåÂøÇÑ Æ÷Áî, ¸ÅĪ ¾Ë°í¸®Áò, ½ºÄµ ³ëÀÌÁî·Î ÀÛ¿ëÇÏ´Â ÁÖº¯ Â÷·®µéÀÇ À¯¹« µî¿¡ µû¶ó¼­ »óÀÌÇÑ ¸ÅĪ °á°ú¸¦ ¸¸µé¾î³¾ ¼ö Àֱ⠶§¹®ÀÌ´Ù. ¹«¼öÈ÷ ¸¹Àº Á¶°ÇµéÀÇ Á¶ÇÕ¿¡ ´ëÇؼ­ ½Ç½Ã°£ ½ºÄµÀ» ¾ò¾î¼­ 3Â÷¿ø Á¡±º Áöµµ¿ÍÀÇ ¸ÅĪ ¼º´ÉÀ» Æò°¡ÇÏ´Â °ÍÀº ½Ã°£°ú ºñ¿ë¸é¿¡¼­ ºñÈ¿À²ÀûÀÌ´Ù. º» ³í¹®Àº Â÷·®À¸·ÎºÎÅÍ ½Ç½Ã°£ ½ºÄµÀ» ¾ò´Â °Í ´ë½Å¿¡ 3Â÷¿ø Á¡±º Áöµµ·ÎºÎÅÍ NDS ¿Í OPSÀÇ 2°¡Áö »ùÇøµ ¹æ¹ýÀ» ÀÌ¿ëÇÏ¿© ´Ù¾çÇÑ Á¶°ÇµéÀ» Á¶ÇÕÇÑ °¡»ó ½ºÄµÀ» ÇÕ¼ºÇÏ°í, °¡»ó ½ºÄµÀ» Á¡±º Áöµµ¿Í ¸ÅĪÇÔÀ¸·Î½á ¸ÅĪ ¼º´ÉÀ» Æò°¡ÇÏ´Â ¹æ¹ýÀ» Á¦¾ÈÇÑ´Ù. ±¹³» ÀÚÀ²ÁÖÇà ½ÇÇèµµ½ÃÀÎ K-City¸¦ ´ë»óÀ¸·Î Á¦ÀÛµÈ 3Â÷¿ø Á¡±º Áöµµ¸¦ ÀÌ¿ëÇÏ¿© NDS¿Í OPS¿¡ ´ëÇÑ °¡»ó ½ºÄµÀ» »ý¼ºÇÏ°í °¢ ¹æ¹ý¿¡ ´ëÇÏ¿© ¸ÅĪ Å×½ºÆ®¸¦ ¼öÇàÇÑ °á°ú, ÇØ´ç »ùÇøµ ¹æ¹ýÀ¸·Î »ý¼ºÇÑ °¡»ó ½ºÄµÀÌ ¹°¸® ½ºÄµÀ» ´ëüÇÏ¿© 3Â÷¿ø Á¡±º ÁöµµÀÇ ¸ÅĪ ¼º´É Æò°¡¿¡ È°¿ëµÉ ¼ö ÀÖÀ½À» È®ÀÎÇÏ¿´´Ù
¿µ¹®³»¿ë
(English Abstract)
3D point maps are used to estimate the location of a vehicle with centimeter-level precision by matching with a LiDAR(Light Detection And Ranging) scan acquired in real-time. However, the 3D point map does not always guarantee an accurate localization for a real-time scan received under various conditions and situations. This is because the real-time scan can produce different matching results depending upon the LiDAR model, mounting pose on the vehicle, matching algorithm, and the surrounding vehicles acting as noise. It is inefficient in terms of time and cost to evaluate matching performance with a 3D point map by obtaining a real-time scan for a combination of many conditions. This study proposed a method to evaluate matching performance by synthesizing various virtual scans using two sampling methods of NDS(Normal Distribution-based Sampling) and OPS(Occupancy Probability-based Sampling) from a 3D point map instead of obtaining the real-time scan and matching the virtual scans to the 3D point map. As a result of creating virtual scans of NDS and OPS using a K-City 3D point map, a domestic autonomous experimental driving city, and performing a matching test, it was confirmed that the virtual scan could be used to evaluate the matching performance of the 3D point map by replacing the physical scan.
Å°¿öµå(Keyword) 3Â÷¿ø Á¡±º Áöµµ   ÀÚÂ÷ À§Ä¡ ÃßÁ¤   ½ºÄµ ¸ÅĪ   °¡»ó ¶óÀÌ´õ ½ºÄµ   ¸ÅĪ ¼º´É Æò°¡   3D point map   localization   scan matching   virtual LiDAR scan   matching quality evaluation  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå