조작 능력이 저하된 장애인 및 환자를 위한 퍼스널 모빌리티에 대한 Q-러닝 제어 알고리즘 적용 = Application of Q-Learning Control Algorithms to Personal Mobility for the Disabled and Patients with Weak Maneuvering Abilities

다국어 초록 (Multilingual Abstract)

Social welfare problems are gradually improving, but many problems still exist. In the case of the elderly and patients with lower extremity disorders, one most difficult factor in their daily life is the problem of restricted moving space that considerably restricts their quality of life. Therefore, this study proposed a medical personal mobility device as an auxiliary equipment for patients and the disabled using convergence technologies in various fields such as electrical, electronics, medical, and mechanical engineering as well as artificial intelligence to overcome these living barriers and implemented a table-based Q-learning algorithm for application on low-specification systems. Moreover, this study proposed the state classification method and improved it using modified Softmax functions to select optimal behaviors, and evaluated its performance.

참고문헌 (Reference)

1 Schulman, J, "Trust Region Policy Optimization"

2 Palmas, A, "The Reinforcement Learning Workshop" Packt Publishing Ltd 483-549, 2020

3 Williams, R. J, "Simple Statistical Gradient-Following Algorithms for Connectionist Reinforcement Learning" 8 : 229-256, 1992

4 Sutton, R. S, "Reinforcement Learning: An Introduction" MIT Press 1998

5 Altuntas, N, "Reinforcement Learning-based Mobile Robot Navigation" 24 (24): 1747-1767, 2016

6 Watkins, C. J. C. H, "Q-Learning" 8 : 279-292, 1992

7 Schulman, J, "Proximal Policy Optimization Algorithms"

8 Mnih, V, "Playing Atari with Deep Reinforcement Learning"

9 Ye, D, "Modeling, Simulation and Fabrication of a Balancing Robot, 2.151: Advanced System Dynamics & Control" Massachusetts Institute of Technology 2012

10 Siegwart, R, "Introduction to Autonomous Mobile Robots" MIT Press 62-103, 2004

11 Geron, A, "Hands-on Machine Learning with Scikit-Learn, Keras and TensorFlow" O’Reilly Media 2019