BROWSE
Detail
Conference Paper
가이드독 로봇과 사용자 간 대형 제어 전략 설계
Cited - time in 
Share 
Share
- Issue Date
- 2024-06
- Citation
- 대한전자공학회 학술 대회 (하계) 2024, pp.1-4
- Publisher
- 대한전자공학회
- Language
- Korean
- Type
- Conference Paper
- Abstract
- This paper presents a design of formation control strategies between a human user and a guide dog robot through reinforcement learning, considering user acceptance. The state space is defined by the surrounding environment observations, and the discrete action space is defined by the relative positions between the user and the guide dog. It includes considerations for designing the value function, addressing trade-offs between the user's walking speed and the ability to overcome narrow paths in different formations. The paper also introduces a simulator environment design for testing various reinforcement learning algorithms and provides an example of the robot system design for conducting experiments in real-world settings.
- KSP Keywords
- Action space, Control strategy, Formation control, Real-world, Reinforcement learning(RL), Relative position, State space, Surrounding environment, Trade-off, User acceptance, learning algorithm