åǥÁö

[GT] ÀÚÀ²ÁÖÇà¿¡ µµ¿òÀÌ µÇ´Â ¹°Ã¼ ½Äº° ½Ã½ºÅÛ °³¹ß

2020³â 6¿ù 16ÀÏ, ÄÄÇ»ÅÍ ºñÀü(Vision) µö·¯´× ºÐ¾ß ÃÖ°í ¼öÁØ ÇÐȸ CVPR(Conference on Computer Vision and Pattern Recognition, ÄÄÇ»ÅÍ ºñÀü ¹× ÆÐÅÏ ÀÎ½Ä ÄÁÆÛ·±½º..



ÀÚÀ²ÁÖÇà¿¡ µµ¿òÀÌ µÇ´Â ¹°Ã¼ ½Äº° ½Ã½ºÅÛ °³¹ß

 

¾î¶² ½Å±â¼úÀÌ ¼¼»óÀ» ±ØÀûÀ¸·Î º¯È­½Ãų±î? ¼¼°è ÃÖ°íÀÇ ¿¬±¸¼Ò¿¡¼­ ³ª¿À´Â ³î¶ó¿î Çõ½ÅÀ» µ¶Á¡ ¼Ò°³ÇÕ´Ï´Ù.

 

2020³â 6¿ù 16ÀÏ, ÄÄÇ»ÅÍ ºñÀü(Vision) µö·¯´× ºÐ¾ß ÃÖ°í ¼öÁØ ÇÐȸ CVPR(Conference on Computer Vision and Pattern Recognition, ÄÄÇ»ÅÍ ºñÀü ¹× ÆÐÅÏ ÀÎ½Ä ÄÁÆÛ·±½º)¿¡¼­ ¹ßÇ¥ µÈ ÇÑ ¿¬±¸´Â ÀÚµ¿Â÷, ÀÚÀü°Å, º¸ÇàÀÚ¸¦ Æ÷ÇÔÇÏ´Â »ç¹°À» ±¸º°ÇÏ°í, ±× ¹æÇâ°ú ´Ù°¡¿À´Â ¼Óµµ¸¦ – ½ÉÁö¾î ¸ðÅüÀÌ¿¡¼­ ³ª¿À´Â »ç¹°±îÁöµµ - ÃøÁ¤ÇÒ ¼ö ÀÖ´Â ½Ã½ºÅÛÀ» ¼Ò°³Çß´Ù. ÀÌ Á¢±Ù ¹æ½ÄÀ» »ç¿ëÇÏ¸é ½ÇÁ¦ ÀÚÀ² ÁÖÇà ½Ã³ª¸®¿À¿¡¼­ ±âÁ¸ÀÇ ÇÑ ¹æÇ⠽𢠼¾¼­¸¦ ÅëÇØ º¸ÇàÀÚ¿Í ÀÚÀü°Å¸¦ Æ÷ÂøÇϱâ Àü¿¡, ÀÌ ½Ã½ºÅÛÀ» ÅëÇØ ±×µé¿¡ ´ëÇÑ »çÀü Ãæµ¹ °æ°í°¡ °¡´ÉÇØÁø´Ù.

 

ÀÌ ¿¬±¸´Â ÇÁ¸°½ºÅÏ ´ëÇп¡¼­ ½ÃÀ۵Ǿú°í, Ãʱ⠿¬±¸¿¡¼­ ¿¬±¸ÆÀÀº ºûÀ» »ç¿ëÇÏ¿© ¸ðÅüÀÌ¿¡ ¼ûÀº »ç¹°À» È®ÀÎÇß´Ù. ±×·¯³ª ÀÌ·¯ÇÑ ½Ãµµ´Â ÇöÀçÀÇ ÀÚµ¿Â÷¿¡ È°¿ëÇϱ⿡´Â ½Ç¿ëÀûÀÌÁö ¾Ê¾Ò´Ù. °íÃâ·Â ·¹ÀÌÀú°¡ ÇÊ¿äÇÏ°í Àû¿ë °Å¸®µµ ª°Ô Á¦ÇѵDZ⠶§¹®ÀÌ´Ù.

 

ÀÌ¿¡ Ãʱ⠿¬±¸¸¦ ¼öÇàÇÑ ÀÌ ¿¬±¸ÆÀÀº °¡½Ã±¤¼± ´ë½Å À̹Ì¡(imaging) ·¹ÀÌ´õ¸¦ »ç¿ëÇÏ¿© ÀÚµ¿Â÷ÀÇ ½Ã¼±¿¡¼­ ¹þ¾î³­ À§ÇèÀ» °¨ÁöÇÏ´Â ½Ã½ºÅÛÀ» ¸¸µé ¼ö ÀÖ´ÂÁö °í¹ÎÇß´Ù. ¸Å²ô·¯¿î Ç¥¸é¿¡¼­ÀÇ ½ÅÈ£ ¼Õ½ÇÀº ·¹ÀÌ´õ ½Ã½ºÅÛÀÇ °æ¿ì ÈξÀ ³·À¸¸ç ·¹ÀÌ´õ´Â ¹°Ã¼ ÃßÀûÀ» À§ÇÑ ÀÔÁõµÈ ±â¼úÀÌ´Ù. ¹®Á¦´Â ·¹ÀÌ´õÀÇ °ø°£ ÇØ»óµµ - ÀÚµ¿Â÷ ¹× ÀÚÀü°Å¿Í °°Àº ¸ðÅüÀÌ ÁÖº¯ÀÇ »ç¹°À» ÃÔ¿µÇÏ´Â µ¥ »ç¿ëµÊ - °¡ »ó´ëÀûÀ¸·Î ³·´Ù´Â °ÍÀÌ´Ù. ÇÏÁö¸¸ ´ÙÇàÈ÷µµ ¿¬±¸ÆÀÀº ¼¾¼­°¡ ÀÛµ¿ÇÒ ¼ö ÀÖµµ·Ï ·¹ÀÌ´õ µ¥ÀÌÅ͸¦ Çؼ®ÇÏ´Â ¾Ë°í¸®ÁòÀ» ¸¸µé¾ú´Ù.

 

À̵éÀÌ °³¹ßÇÑ ¾Ë°í¸®ÁòÀº ¸Å¿ì È¿À²ÀûÀ̸ç Çö ¼¼´ëÀÇ ÀÚµ¿Â÷ Çϵå¿þ¾î ½Ã½ºÅÛ¿¡ ÀûÇÕÇÏ´Ù. µû¶ó¼­ ¿ì¸®´Â Â÷¼¼´ë Â÷·®¿¡¼­ ÀÌ ±â¼úÀ» º¼ ¼ö ÀÖÀ»Áöµµ ¸ð¸¥´Ù.

 

½Ã½ºÅÛÀÌ »ç¹°À» ±¸º°ÇÒ ¼ö ÀÖµµ·Ï ¿¬±¸ÆÀÀº Ç¥ÁØ ·¹ÀÌ´õ°¡ À¯¿ëÇÑ Á¤º¸°¡ ¾Æ´Ñ ¹è°æ ¼ÒÀ½À¸·Î °£ÁÖÇÏ´Â ·¹ÀÌ´õ ½ÅÈ£ÀÇ ÀϺθ¦ ó¸®Çß°í, ÀΰøÁö´É ±â¼úÀ» Àû¿ëÇÏ¿© µ¿ÀÏ ÇÁ·Î¼¼½º¸¦ °³¼±ÇÏ°í À̹ÌÁö¸¦ ÀÐÀ» ¼ö ÀÖµµ·Ï Çß´Ù. À̸¦ À§ÇØ ½Ã½ºÅÛÀ» ½ÇÇàÇÏ´Â ÄÄÇ»ÅÍ´Â ¸Å¿ì ÀûÀº ¾çÀÇ µ¥ÀÌÅÍ·Î ÀÚÀü°Å¿Í º¸ÇàÀÚ¸¦ ÀνÄÇÏ´Â ¹æ¹ýÀ» ÇнÀÇØ¾ß Çß´Ù.

 

¿ì¼±, ¹«¾ùÀÌ ±×°÷¿¡ ÀÖ´ÂÁö¸¦ °¨ÁöÇØ¾ß ÇÑ´Ù. ±×¸®°í ¹«¾ùÀÌ ±×°÷¿¡ ÀÖÀ¸¸é, ½Ã½ºÅÛÀº ¡®±×°ÍÀÌ Áß¿äÇմϱî?¡¯¶ó°í ¹¯°Ô µÈ´Ù. Áï, ¡®ÀÚÀü°ÅÀԴϱî, º¸ÇàÀÚÀԴϱî?¡¯¸¦ ¹¯°í, ÀÌÈÄ ±×°ÍÀÇ Á¤È®ÇÑ À§Ä¡¸¦ ã¾Æ³»¾ß ÇÑ´Ù.

 

ÀÌ ½Ã½ºÅÛÀº ÇöÀç º¸ÇàÀÚ¿Í ÀÚÀü°Å¸¦ °¨ÁöÇÏ´Â µ¥ ¿ì¼±¼øÀ§¸¦ µÎ´Âµ¥, ÀÌ´Â ¿£Áö´Ï¾îµéÀÌ ±×·¯ÇÑ °ÍµéÀÌ °¡Àå µµÀüÀûÀÌ°í ±î´Ù·Î¿î °úÁ¦¶ó°í ¿©±â±â ¶§¹®ÀÌ´Ù. ¿ì¼± Å©±â°¡ ÀÛ°í, ¾î¶² ƯÁ¤ Ʋ¿¡ ¾ô¸ÅÀÌÁö ¾Ê´Â ¸ð¾çÀ» Áö´Ï¸ç µ¿ÀÛ ¶ÇÇÑ ´Ù¾çÇϱ⠶§¹®ÀÌ´Ù. À̸¦ °³¼±Çϸé, ÀÚµ¿Â÷´Â ½±°Ô °¨ÁöµÇµµ·Ï ½Ã½ºÅÛÀÌ Á¶Á¤µÉ ¼ö ÀÖ´Ù.

 

¿¬±¸ÆÀÀº ·¹ÀÌ´õ¿Í ½ÅÈ£ ó¸® °³¼±À» Æ÷ÇÔÇÏ´Â ´Ù°¢µµÀÇ Àû¿ë ¹æ¾È¿¡ ´ëÇÑ ¿¬±¸¸¦ ÁøÇàÇÒ °èȹÀÌ´Ù. ÀÌ ½Ã½ºÅÛÀº ÀÚµ¿Â÷ ¾ÈÀüÀ» ȹ±âÀûÀ¸·Î Çâ»ó½Ãų ¼ö ÀÖ´Â ÀáÀç·ÂÀ» °¡Áö°í ÀÖ´Ù. ¹°·Ð »ó¿ëÈ­ ÀÌÀü¿¡, ÀÌ ±â¼úÀº ¸Å¿ì ¾ö°ÝÇÑ ÀÚµ¿Â÷°³¹ß Áֱ⸦ °ÅÃľßÇÑ´Ù. °¢ ±â±â¿Í ÅëÇÕµÇ°í ½ÃÀå¿¡ Ãâ½ÃÇϱâ±îÁö, ¿©ÀüÈ÷ ¸¹Àº ¿£Áö´Ï¾î¸µ °úÁ¤À» ÇÊ¿ä·Î ÇÑ´Ù. ±×·¯³ª ¿øõ ±â¼úÀÌ È®º¸µÇ¾ú±â ¶§¹®¿¡, ¿ì¸®°¡ °ð ÀÚµ¿Â÷¿¡¼­ ÀÌ ±â¼ú°ú ¸¸³ªº¼ ¼ö ÀÖÀ» È®·üÀÌ ³ô´Ù.

 

- Princeton University, Engineering School, June 24, 2020, ¡°New Radar Lets Cars Spot Hazards Around Corners,¡± by John Sullivan.  © 2020 The Trustees of Princeton University.  All rights reserved.

 

- To view this article, please visit:
https://engineering.princeton.edu/news/2020/06/24/new-radar-lets-cars-spot-hazards-around-corners

What new technologies will dramatically transform your world?  We¡¯ll present an exclusive preview of the stunning breakthroughs emerging from the world¡¯s leading research labs.

 

**

 

Research presented June 16 at the Conference on Computer Vision and Pattern Recognition (or CVPR), described a system that is able to distinguish objects including cars, bicyclists, and pedestrians and gauge their direction and oncoming speed, even around corners. This approach allows for collision warning for pedestrians and cyclists in real-world autonomous driving scenarios - before seeing them with existing direct line-of-sight sensors.

 

In earlier research, the Princeton-based team used light to see objects hidden around corners. But those efforts are currently not practical for use in cars both because they require high-powered lasers and because they are restricted to short ranges.

 

While conducting that earlier research, the team wondered whether it would be possible to create a system to detect hazards out of the car¡¯s line of sight using imaging radar instead of visible light. The signal loss at smooth surfaces is much lower for radar systems, and radar is a proven technology for tracking objects. The challenge is that radar¡¯s spatial resolution - used for picturing objects around corners such as cars and bikes - is relatively low. But fortunately, the researchers created algorithms to interpret the radar data allowing the sensors to function.

 

The algorithms that they developed are highly efficient and fit on current generation automotive hardware systems.  So, we might see this technology in the next generation of vehicles.

 

To allow the system to distinguish objects, the team processed a part of the radar signal that standard radars consider background noise rather than useful information. The team applied artificial intelligence techniques to refine the processing and read the images. To do this, the computer running the system had to learn to recognize cyclists and pedestrians from a very sparse amount of data.

 

First, it must detect if something is there. If there is something there, it asks ¡°is it important?¡± That is, ¡°Is it a cyclist or a pedestrian?¡±  Then it must locate it.


The system currently prioritizes detecting pedestrians and cyclists because the engineers feel those are the most challenging objects, because of their small size, unconventional shapes, and varied motion. Once that is refined, the system will be adjusted to detect vehicles as well.

 

The researchers plan to follow the research in a number of directions for applications involving both radar and refinements in signal processing. This system has the potential to radically improve automotive safety.

 

Before becoming commercialized, it will have to go through very rigorous automotive development cycles. In terms of integration and bringing it to market, it still requires a lot of engineering. But the technology is there, so there is the potential for seeing this in vehicles very soon.¡±

 

References
Princeton University, Engineering School, June 24, 2020, ¡°New Radar Lets Cars Spot Hazards Around Corners,¡± by John Sullivan.  © 2020 The Trustees of Princeton University.  All rights reserved.


To view this article, please visit:

https://engineering.princeton.edu/news/2020/06/24/new-radar-lets-cars-spot-hazards-around-corners