Download PDFOpen PDF in browser

Reinforcement Learning of Active Vision for Object Manipulation Under Occlusions

EasyChair Preprint no. 11784

7 pagesDate: January 17, 2024


This research presents a novel approach to enhance the capabilities of robotic systems in object manipulation tasks under occlusions through the application of reinforcement learning (RL) techniques. Occlusions pose a significant challenge in robotic manipulation scenarios, as they hinder the robot's ability to perceive and interact with objects effectively. The RL model is trained to make intelligent decisions on when and where to move the robot's sensors actively to gather informative visual cues, enabling effective object recognition and manipulation planning even in the presence of occlusions. The learning process is driven by a reward mechanism that encourages the robot to explore and adapt its vision-based strategies, ultimately improving its ability to handle occluded scenarios. This research contributes valuable insights into the application of RL for addressing challenges in robotic manipulation, particularly in scenarios involving occluded objects.

Keyphrases: Effective, Robotic, vision

BibTeX entry
BibTeX does not have the right entry for preprints. This is a hack for producing the correct reference:
  author = {Kurez Oroy and Jane Smith},
  title = {Reinforcement Learning of Active Vision for Object Manipulation Under Occlusions},
  howpublished = {EasyChair Preprint no. 11784},

  year = {EasyChair, 2024}}
Download PDFOpen PDF in browser