Á¤º¸Ã³¸®ÇÐȸ ³í¹®Áö ÄÄÇ»ÅÍ ¹× Åë½Å½Ã½ºÅÛ
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
µö·¯´× ±â¹Ý °´Ã¼ Àνİú ÃÖÀû °æ·Î Ž»öÀ» ÅëÇÑ ¸ÖƼ Àç³ µå·Ð ½Ã½ºÅÛ ¼³°è ¹× ±¸Çö¿¡ ´ëÇÑ ¿¬±¸ |
¿µ¹®Á¦¸ñ(English Title) |
A Study on the Design and Implementation of Multi-Disaster Drone System Using Deep Learning-Based Object Recognition and Optimal Path Planning |
ÀúÀÚ(Author) |
±èÁøÇõ
ÀÌÅÂÈñ
º¯ÈñÁ¤
Jin-Hyeok Kim
Tae-Hui Lee
Yamin Han
Heejung Byun
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 10 NO. 04 PP. 0117 ~ 0122 (2021. 04) |
Çѱ۳»¿ë (Korean Abstract) |
ÃÖ±Ù ÅÂdz, ÁöÁø, »êºÒ, »ê»çÅÂ, ÀüÀï µî ´Ù¾çÇÑ Àç³ »óȲÀ¸·Î ÀÎÇÑ ÀθíÇÇÇØ¿Í ÀÚ±Ý ¼Õ½ÇÀÌ ²ÙÁØÈ÷ ¹ß»ýÇÏ°í ÀÖ°í ÇöÀç À̸¦ ¿¹¹æÇÏ°í º¹±¸Çϱâ À§ÇØ ¸¹Àº Àη°ú ÀÚ±ÝÀÌ ¼Ò¿äµÇ°í ÀÖ´Â ½ÇÁ¤ÀÌ´Ù. ÀÌ·¯ÇÑ ¿©·¯ Àç³ »óȲÀ» ¹Ì¸® °¨½ÃÇÏ°í Àç³ ¹ß»ýÀÇ ºü¸¥ ÀÎÁö ¹× ´ëó¸¦ À§ÇØ º» ³í¹®¿¡¼´Â ÀΰøÁö´É ±â¹ÝÀÇ Àç³ µå·Ð ½Ã½ºÅÛÀ» ¼³°è ¹× °³¹ßÇÏ¿´´Ù. º» ¿¬±¸¿¡¼´Â »ç¶÷ÀÌ °¨½ÃÇϱâ Èûµç Áö¿ª¿¡ ¿©·¯ ´ëÀÇ Àç³ µå·ÐÀ» ÀÌ¿ëÇÏ¸ç µö·¯´× ±â¹ÝÀÇ °´Ã¼ ÀÎ½Ä ¾Ë°í¸®Áò°ú ÃÖÀû °æ·Î Ž»ö ¾Ë°í¸®ÁòÀ» Àû¿ëÇØ °¢°¢ÀÇ µå·ÐÀÌ ÃÖÀûÀÇ °æ·Î·Î È¿À²Àû Ž»öÀ» ½Ç½ÃÇÑ´Ù. ¶ÇÇÑ µå·ÐÀÇ ±Ùº»Àû ¹®Á¦ÀÎ ¹èÅ͸® ¿ë·® ºÎÁ·¿¡ ´ëÇÑ ¹®Á¦Á¡À» ÇØ°áÇϱâ À§ÇØ Ant Colony Optimization (ACO) ±â¼úÀ» ÀÌ¿ëÇÏ¿© °¢ µå·ÐÀÇ ÃÖÀû °æ·Î¸¦ °áÁ¤ÇÏ°Ô µÈ´Ù. Á¦¾ÈÇÑ ½Ã½ºÅÛ ±¸ÇöÀ» À§ÇØ ¿©·¯ Àç³ »óȲ Áß »êºÒ »óȲ¿¡ Àû¿ëÇÏ¿´À¸¸ç Àü¼ÛµÈ µ¥ÀÌÅ͸¦ ±â¹ÝÀ¸·Î »êºÒÁöµµ¸¦ ¸¸µé°í, ºö ÇÁ·ÎÁ§Å͸¦ žÀçÇÑ µå·ÐÀÌ Ã⵿ÇÑ ¼Ò¹æ°ü¿¡°Ô »êºÒÁöµµ¸¦ ½Ã°¢ÀûÀ¸·Î º¸¿©ÁÖ¾ú´Ù. Á¦¾ÈÇÑ ½Ã½ºÅÛ¿¡¼´Â ¿©·¯ ´ëÀÇ µå·ÐÀÌ ÃÖÀû °æ·Î Ž»ö ¹× °´Ã¼ÀνÄÀ» µ¿½Ã¿¡ ¼öÇàÇÔÀ¸·Î½á ºü¸¥ ½Ã°£ ³»¿¡ Àç³ »óȲÀ» ÀÎÁöÇÒ ¼ö ÀÖ´Ù. º» ¿¬±¸¸¦ ¹ÙÅÁÀ¸·Î Àç³ µå·Ð ÀÎÇÁ¶ó¸¦ ±¸ÃàÇÏ°í Á¶³ÀÚ Å½»ö(¹Ù´Ù, »ê, ¹Ð¸²), µå·ÐÀ» ÀÌ¿ëÇÑ ÀÚüÀûÀÎ ÈÀçÁø¾Ð, ¹æ¹ü µå·Ð µî¿¡ È°¿ëÇÒ ¼ö ÀÖ´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
In recent years, human damage and loss of money due to various disasters such as typhoons, earthquakes, forest fires, landslides, and wars are steadily occurring, and a lot of manpower and funds are required to prevent and recover them. In this paper, we designed and developed a disaster drone system based on artificial intelligence in order to monitor these various disaster situations in advance and to quickly recognize and respond to disaster occurrence. In this study, multiple disaster drones are used in areas where it is difficult for humans to monitor, and each drone performs an efficient search with an optimal path by applying a deep learning-based optimal path algorithm. In addition, in order to solve the problem of insufficient battery capacity, which is a fundamental problem of drones, the optimal route of each drone is determined using Ant Colony Optimization (ACO) technology. In order to implement the proposed system, it was applied to a forest fire situation among various disaster situations, and a forest fire map was created based on the transmitted data, and a forest fire map was visually shown to the fire fighters dispatched by a drone equipped with a beam projector. In the proposed system, multiple drones can detect a disaster situation in a short time by simultaneously performing optimal path search and object recognition. Based on this research, it can be used to build disaster drone infrastructure, search for victims (sea, mountain, jungle), self-extinguishing fire using drones, and security drones.
|
Å°¿öµå(Keyword) |
±ºÁý
µå·Ð
µö·¯´×
°æ·Î ¼³Á¤
ACO
Swarm
Drone
Deep Learning
Path Planning
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|