Deep reinforcement learning for cooperative robots based on adaptive sentiment feedback

Haein Jeon, Dae-Won Kim, Bo-Yeong Kang

Abstract

Highlights•A robot teaching strategy for cooperative tasks via human–robot interaction is proposed.•Deep Q-Network is employed with real-time feedback delivered by the trainer’s speech.•A novel reward function is designed to guide the robot in natural interactions.•The function incorporates the feedback with improved reward-shaping techniques.•Experiments on a robot proved the effectiveness of the proposed approach.
BibTeX
@article{jeon2024deep,
  author = {Haein Jeon and Dae-Won Kim and Bo-Yeong Kang},
  title = {{Deep reinforcement learning for cooperative robots based on adaptive sentiment feedback}},
  journal = {Expert Systems with Applications},
  year = {2024},
  doi = {10.1016/j.eswa.2023.121198},
}

Loading related content...

Figure