ニューラルネットワークを用いたペンのはめ合いタスクにおける力情報による成否判定  [in Japanese] Success Detection by Force Information in Fitting Task of Pen Using Neural Network  [in Japanese]

Access this Article

Author(s)

Abstract

<p>This paper proposes a success detection system for tasks with high generalization performance by neural networks using force information. Automation of anomaly detection is important. Especially, accurate judgment of the success and failure of tasks is important. For example recognizing the failure of the task improves the reliability of the task . However, conventional methods using machine learning cannot judge success and failure for objects which has not been trained. This paper showed that the neural network using force information can judge success and failure of tasks for objects which cannot be trained. This paper compared success detection methods using force information, image information, or position information. The results showed the method with force information resulted in the highest generalitation objects in a pen assembly task.</p>

Journal

  • The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec)

    The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec) 2019(0), 2P1-R04, 2019

    The Japan Society of Mechanical Engineers

Codes

Page Top