๐Ÿ”๏ธ

SLAM

Tags
Self-driving Car
ID matched
Created
Jul 13, 2023 12:43 PM
Last Updated
Last updated July 15, 2023
ย 
ย 
ย 

SLAM

  • Simultaneous Localization and Mapping
  • ๋™์‹œ์  ์œ„์น˜์ถ”์ • ๋ฐ ์ง€๋„ ์ž‘์„ฑ
  • A robot
    • Estimates its motion
    • Builds a model of its surroundings
    • Using a specific sensor
    • Without a prior information
  • When?
    • Low-quality map exists
    • Low-quality position information exists
    • .. or none of them exists
  • Does what?
    • Estimates optimal pose + map information, without any prior information
ย 
ย 

Mobile robotics

  • ์ด๋™ ๊ฐ€๋Šฅํ•œ ๋กœ๋ณดํ‹ฑ์Šค
    • ์‚ฌ๋žŒ์ด ๊ฐ€์ง€ ๋ชปํ•˜๋Š” ๊ณณ์„ ๋Œ€์‹  ํƒ์ƒ‰
      • ์œ„ํ—˜ํ•œ ๊ณณ
      • ์‚ฌ๋žŒ์ด ๊ฐ€๊ธฐ ์–ด๋ ค์šฐ ๊ณณ / ์ข์€ ๊ณณ
    • ์ด๋™ ์ž์ฒด๋ฅผ ์ž๋™ํ™”
      • ๋น„์šฉ - ์‚ฌ๋žŒ vs ๋กœ๋ด‡ ํˆฌ์ž… ๋น„์šฉ
      • ์šด์ „์„ ์ž๋™ํ™”
  • ๊ณ ์ •๋œ ๋กœ๋ด‡์˜ workspace ํ™•์žฅ
  • ์ž์œจ์ด๋™์ฒด์˜ ์กฐ๊ฑด (์ธ์ง€, ๊ฒฐ์ •, ํ–‰๋™)
    • Perception (์ฃผ๋ณ€ ๊ณต๊ฐ„ ์ธ์ง€)
    • Exteroceptive sensing
    • Proprioceptive sensing
ย 
ย 

Localization & Mapping

  • Localization
    • Get optimal value from low-quality exteroceptive sensing, using hints from high-quality proprioceptive sensing
    • Given accurate motion information
    • Reconstruct a map
    • When?
      • High-quality map exists
    • Does what?
      • Estimates optimal pose, with prior map
  • Mapping
    • Get optimal value from low-quality proprioceptive, using hints from high-quality exteroceptive sensing
    • Given an accurate map
    • Estimate environment
    • When?
      • High-quality position information exists
    • Does what?
      • Estimates optimal map, with prior pose information
ย 
ย 

Sensors

  • Wheel encoder
    • ๋ฐ”ํ€ด์˜ ํšŒ์ „๋Ÿ‰๊ณผ ์ด๋™๋Ÿ‰์„ ์ธก์ •ํ•˜๋Š” ์„ผ์„œ
  • IMU
    • Linear accelerator ์„ผ์„œ์™€ Angular gyroscope ์„ผ์„œ๊ฐ€ ํ˜ผํ•ฉ๋œ ์„ผ์„œ
  • GNSS (Global navigation satellite system)
    • ๋น„์ฝ˜ ๊ธฐ๋ฐ˜์˜ ์œ„์น˜ ์ถ”์ • ์„ผ์„œ
  • LiDAR
    • ์ ์™ธ์„  ๋ ˆ์ด์ €๋ฅผ ์˜๊ณ  ๋ฐ˜์‚ฌ ์‹œ๊ฐ„์„ ์ธก์ •ํ•˜์—ฌ ๊ฑฐ๋ฆฌ๋ฅผ ์ถ”์ •ํ•˜๋Š” ์„ผ์„œ
  • RADAR
    • ๋ฐ˜์‚ฌ๋˜์–ด ๋Œ์•„์˜ค๋Š” ์ „ํŒŒ๋ฅผ ์ธก์ •ํ•˜์—ฌ radial ๊ฑฐ๋ฆฌ๋ฅผ ์žฌ๋Š” ์„ผ์„œ
  • Ultrasound
    • ์ดˆ์ŒํŒŒ๋ฅผ ์ด์šฉ - RADAR์™€ ์ž‘๋™ ๋ฐฉ์‹ ๋™์ผ
  • Camera
    • ๊ด‘์„ผ์„œ๋ฅผ ์ด์šฉํ•ด ๋น› ์‹ ํ˜ธ๋ฅผ ๋ฐ›๊ณ , debayering ํ”„๋กœ์„ธ์Šค๋ฅผ ํ†ตํ•ด RGB ์ƒ‰ ์žฌ๊ตฌ์„ฑ
  • Microphones
    • ๊ณต๊ธฐ์˜ ์ง„๋™์„ transducer ์„ผ์„œ๋ฅผ ํ†ตํ•ด ์ „๊ธฐ ์‹ ํ˜ธ๋กœ ๋ณ€ํ™˜ํ•˜๋Š” ์„ผ์„œ
    • ์—ฌ๋Ÿฌ๊ฐœ์˜ ๋งˆ์ดํฌ๋ฅผ ํ†ตํ•ด ์†Œ๋ฆฌ์˜ ๊ทผ์›์— ๋Œ€ํ•œ ์œ„์น˜๋ฅผ ๊ณ„์‚ฐ ๊ฐ€๋Šฅ
ย 
ย