A Closed-Loop Multi-perspective Visual Servoing Approach with Reinforcement Learning

IEEE International Conference on Robotics and Biomimetics (ROBIO), pp. 1–7

Abstract

Traditional visual servoing methods suffer from serving between scenes from multiple perspectives, which humans can complete with visual signals alone. In this paper, we investigated how multi-perspective visual servoing could be solved under robot-specific constraints, including self-collision, singularity problems. We presented a novel learning-based multi-perspective visual servoing framework, which iteratively estimates robot actions from latent space representations of visual states using reinforcement learning. Furthermore, our approaches were trained and validated in a Gazebo simulation environment with connection to OpenAI/Gym. Through simulation experiments, we showed that our method can successfully learn an optimal control policy given initial images from different perspectives, and it outperformed the Direct Visual Servoing algorithm with mean success rate of 97.0%.

Authors 4

1. Jiacheng Pei Aachen

   RWTH Aachen University

   Affiliation as printed

   Agile Robots AG

   RWTH Aachen University

2. Kaixin Bai

   Universität Hamburg

   Affiliation as printed

   University of Hamburg,TAMS (Technical Aspects of Multimodal Systems),Department of Informatics

   Agile Robots AG

   Department of Informatics, TAMS (Technical Aspects of Multimodal Systems), University of Hamburg

3. Zhaopeng Chen

   Universität Hamburg

   Affiliation as printed

   Agile Robots AG

   Department of Informatics, TAMS (Technical Aspects of Multimodal Systems), University of Hamburg

4. Jianwei Zhang

   Universität Hamburg

   Affiliation as printed

   University of Hamburg,TAMS (Technical Aspects of Multimodal Systems),Department of Informatics

   Department of Informatics, TAMS (Technical Aspects of Multimodal Systems), University of Hamburg