다중 센서 융합을 이용한 로봇 매니퓰레이터의 모멘텀 충돌 관측기

정다운; 김문유; 정주노

doi:10.5302/J.ICROS.2020.20.0154

주제분류

...

저널정보

제어로봇시스템학회
제어로봇시스템학회 논문지 학술저널
제어로봇시스템학회 논문지 제26권 제12호
2020.12 1,054 - 1,061 (8page)
DOI : 10.5302/J.ICROS.2020.20.0154

저자정보

정다운 (고려대학교)
김문유 (고려대학교)
정주노 (고려대학교)

이용수
내서재: 0

내서재에 추가
되었습니다. 내서재에서
삭제되었습니다.

초록·키워드

오류제보하기

In this paper, we propose a multi-sensor-based collision detection algorithm for free and contact situations. The important characteristics of this collision detection algorithm include correctness, fast detection, and versatility. An appropriate combination of suitable multiple sensors such as encoders, force/torque sensors, and accelerometers can be used to improve the proposed collision detection algorithm in terms of these characteristics. Moreover, the dynamic parameters of a robot manipulator must be modeled accurately to obtain better estimates of any possible torque deviation caused by a collision. We employ the torque separation technique that allows us to estimate the dynamic parameters separately, and thus, to increase the robustness of parameter estimation. A momentum-residual-based collision observer with additional information from a force/torque sensor and a set of accelerometers is designed to enhance collision detection. When a robot manipulator holds an object, the effect of the object on the robot dynamics should be considered by using force/torque sensor information. The reactivity of collision detection time and the identifiability of blind spots and collision directions are enhanced using cheap accelerometers. The proposed algorithm is verified by employing a six-axis industrial robot manipulator.

#robot manipulator #collision detection #dynamic parameter identification #multi-sensor fusion