저가의 거리센서를 이용한 이동로봇의 자율주행
Autonomous Navigation of a Mobile Robot Using Low-Cost Range Sensors
- 주제(키워드) 이동로봇 , 저가센서 , 자율주행
- 발행기관 고려대학교 대학원
- 지도교수 송재복
- 발행년도 2010
- 학위수여년월 2010. 2
- 학위구분 석사
- 학과 일반대학원 메카트로닉스협동과정
- 세부전공 메카트로닉스전공
- 원문페이지 79 p
- 실제URI http://www.dcollection.net/handler/korea/000000022338
- 본문언어 한국어
- 제출원본 000045592578
초록/요약
Localization, obstacle avoidance and mapping are fundamental and important for mobile robot navigation. These techniques have been usually implemented by a laser scanner, but its high cost prevents it from being widely used for service robots at home and office environments. Therefore, low-cost range sensors are preferred practically, but they suffer from inaccurate and insufficient measurements. This thesis proposes three novel approaches for low-cost sensor based navigation of a mobile robot. First, a method to increase the success rate of low-cost sensor-based localization is proposed. Instead of increasing the number of sensors, both the previous and the current data obtained from the IR sensors are used for localization in order to utilize as much environmental information as possible. The sensor model used in the monte carlo localization (MCL) is modified so that the accumulated range information can be used to increase the accuracy of pose estimation. The experimental results show that the proposed method can robustly estimate the robot pose in indoor environments with several similar places. Second, this thesis proposes a method to avoid obstacles using low-cost anisotropic ultrasonic sensors with wide beam angle. An ultrasonic sensor has been widely used as a range sensor because of its low cost and capacity of detecting obstacles which are not well detected by a laser scanner and an IR sensor. In this study, only one transmitter and three receivers were used and most obstacles can be detected by the proposed configuration of sensors. Also, this study exploits the probabilistic mapping technique to avoid collision with various types of obstacles. Therefore, the robot can avoid wide obstacles as well as narrow obstacles successfully. The experimental results show that the proposed method can robustly avoid obstacles in most indoor environments. Third, a method to build a grid map using ultrasonic sensors is proposed. However, the grid map constructed by ultrasonic sensors cannot represent the real environment well due to incorrect measurements such as specular reflection. In this thesis, reliability of measurements from ultrasonic sensors was considered for map building and the data association was introduced to estimate reliability of measurements from ultrasonic sensors. The experimental results show that the proposed mapping method is able to build an accurate map for navigation.
more목차
제1장 서론 1
1.1 연구 배경 1
1.2 연구 동향 1
1.3 연구 목표 3
제2장 위치추정 5
2.1 적외선센서의 특성 및 구성 5
2.2 누적된 거리정보를 이용한 가상의 적외선센서 6
2.3 몬테카를로 위치추정 11
2.4 이동모델과 센서모델 14
2.5 위치추정 실험 및 결론 17
제3장 장애물 회피 22
3.1 이방성 초음파센서의 특성 및 구성 22
3.2 장애물의 위치측정 23
3.3 확률기반의 국부지도를 이용한 장애물 표현 27
3.4 Dynamic Window Approach 32
3.5 장애물 회피 실험 및 결론 34
제4장 지도작성 39
4.1 초음파센서의 특성 및 구성 39
4.2 초음파 거리정보의 교합과 형상인지지수 41
4.3 초음파 데이터의 신뢰도를 고려한 격자지도 작성 49
4.4 초음파 격자지도의 평가 및 결론 58
제5장 결 론 67
참고 문헌 68
