Á¤º¸Ã³¸®ÇÐȸ ³í¹®Áö ¼ÒÇÁÆ®¿þ¾î ¹× µ¥ÀÌÅÍ °øÇÐ
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
RGB-D ȯ°æÀÎ½Ä ½Ã°¢ Áö´É, ¸ñÇ¥ »ç¹° °æ·Î Ž»ö ¹× ½ÉÃþ °ÈÇнÀ¿¡ ±â¹ÝÇÑ »ç¶÷Çü ·Îº¿¼ÕÀÇ ¸ñÇ¥ »ç¹° ÆÄÁö |
¿µ¹®Á¦¸ñ(English Title) |
Grasping a Target Object in Clutter with an Anthropomorphic Robot Hand via RGB-D Vision Intelligence, Target Path Planning and Deep Reinforcement Learning |
ÀúÀÚ(Author) |
Ga Hyeon Ryu
Ji-Heon Oh
Jin Gyun Jeong
Hwanseok Jung
Jin Hyuk Lee
Patricio Rivera Lopez
Tae-Seong Kim
·ù°¡Çö
¿ÀÁöÇå
Á¤Áø±Õ
Á¤È¯¼®
ÀÌÁøÇõ
Patricio Rivera Lopez
±èżº
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 11 NO. 09 PP. 0363 ~ 0370 (2022. 09) |
Çѱ۳»¿ë (Korean Abstract) |
´ÙÁß »ç¹° ȯ°æ¿¡¼ ¸ñÇ¥ »ç¹°¸¸ÀÇ Á¤¹ÐÇÑ ÆÄÁö¸¦ À§Çؼ´Â Àå¾Ö¹°°úÀÇ Ãæµ¹ ȸÇÇ Áö´É°ú Á¤±³ÇÑ ÆÄÁö Áö´ÉÀÌ ÇÊ¿äÇÏ´Ù. ÀÌ ÀÛ¾÷À» À§Çؼ± ´ÙÁß »ç¹° ȯ°æ ÀÎÁö, ¸ñÇ¥ »ç¹° ÀνÄ, °æ·Î ¼³Á¤, ·Îº¿¼ÕÀÇ »ç¹° ÆÄÁö Áö´ÉÀÌ ÇÊ¿äÇÏ´Ù. º» ¿¬±¸¿¡¼´Â RGB-D ¿µ»ó ¼¾¼¸¦ ÀÌ¿ëÇÏ¿© ´ÙÁß »ç¹° ȯ°æ°ú »ç¹°À» ÀÎÁöÇÏ°í 3D °ø°£À» ¸ÅÇÎÇÑ ÈÄ, Ãæµ¹ ȸÇÇ °æ·Î Ž»ö ¾Ë°í¸®ÁòÀ» È°¿ëÇÏ¿© ¸ñÇ¥ »ç¹°±îÁöÀÇ °æ·Î¸¦ Ž»ö ¹× ¼³Á¤ÇÏ°í, °ÈÇнÀÀ» ÅëÇØ ÇнÀµÈ »ç¶÷Çü ·Îº¿¼ÕÀÇ ¸ñÇ¥ »ç¹° ÆÄÁö Áö´ÉÀ» È°¿ëÇØ ÃÖÁ¾ÀûÀ¸·Î ½Ã¹Ä·¹ÀÌ¼Ç ¹× Çϵå¿þ¾î »ç¹° ÆÄÁö ½Ã½ºÅÛÀ» ±¸ÇöÇÏ°í °ËÁõÇÏ¿´´Ù. »ç¶÷Çü ·Îº¿¼ÕÀ» ±¸ÇöÇÑ ½Ã¹Ä·¹ÀÌ¼Ç È¯°æ¿¡¼ 5°³ÀÇ »ç¹° Áß ¸ñÇ¥ »ç¹°À» ÁöÁ¤ÇÏ°í ÆÄÁöÇÑ °á°ú °æ·Î Ž»ö ¾ø´Â ÆÄÁö ½Ã½ºÅÛÀÌ Æò±Õ 78.8%ÀÇ ¼º°ø·ü°ú 34%ÀÇ Ãæµ¹·üÀ» º¸ÀÏ ¶§, °æ·Î Ž»ö Áö´É°ú °áÇÕµÈ ½Ã½ºÅÛÀº Æò±Õ 94%ÀÇ ¼º°ø·ü°ú Æò±Õ 20%ÀÇ Ãæµ¹·üÀ» º¸¿´´Ù. UR3¿Í QB-Soft Hand¸¦ »ç¿ëÇÑ Çϵå¿þ¾î ȯ°æ¿¡¼´Â 3°³ÀÇ »ç¹° Áß ¸ñÇ¥ »ç¹°À» ÁöÁ¤ÇÏ°í ÆÄÁöÇÑ °á°ú °æ·Î Ž»ö ¾ø´Â ÆÄÁö ½Ã½ºÅÛÀÌ Æò±Õ 30%ÀÇ ¼º°ø·ü°ú 97%ÀÇ Ãæµ¹·üÀ» º¸ÀÏ ¶§, °æ·Î Ž»ö Áö´É°ú °áÇÕµÈ ½Ã½ºÅÛÀº Æò±Õ 90%ÀÇ ¼º°ø·ü°ú Æò±Õ 23%ÀÇ Ãæµ¹·üÀ» º¸¿´´Ù. º» ¿¬±¸¿¡¼´Â RGB-D ½Ã°¢ Áö´É, Ãæµ¹ ȸÇÇ °æ·Î Ž»ö, »ç¹° ÆÄÁö ½ÉÃþ °ÈÇнÀ Áö´ÉÀÇ °áÇÕÀ» ÅëÇÏ¿©, »ç¶÷Çü ·Îº¿¼ÕÀÇ ¸ñÇ¥ »ç¹° ÆÄÁö°¡ °¡´ÉÇÔÀ» Á¦½ÃÇÏ¿´´Ù. |
¿µ¹®³»¿ë (English Abstract) |
Grasping a target object among clutter objects without collision requires machine intelligence. Machine intelligence includes environment recognition, target & obstacle recognition, collision-free path planning, and object grasping intelligence of robot hands. In this work, we implement such system in simulation and hardware to grasp a target object without collision. We use a RGB-D image sensor to recognize the environment and objects. Various path-finding algorithms been implemented and tested to find collision-free paths. Finally for an anthropomorphic robot hand, object grasping intelligence is learned through deep reinforcement learning. In our simulation environment, grasping a target out of five clutter objects, showed an average success rate of 78.8%and a collision rate of 34% without path planning. Whereas our system combined with path planning showed an average success rate of 94% and an average collision rate of 20%. In our hardware environment grasping a target out of three clutter objects showed an average success rate of 30% and a collision rate of 97% without path planning whereas our system combined with path planning showed an average success rate of 90% and an average collision rate of 23%. Our results show that grasping a target object in clutter is feasible with vision intelligence, path planning, and deep RL. |
Å°¿öµå(Keyword) |
Anthropomorphic Robot Hand
Reinforcement Learning
Path Planning
Object Detection
Path Planning
Object Detection
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|