September 10, 2025ElectronicsOpen Access

Hierarchical Learning for Closed-Loop Robotic Manipulation in Cluttered Scenes via Depth Vision, Reinforcement Learning, and Behaviour Cloning

View Full Paper

Ask AI

Bookmark

Authors

Discussion

This framework improves grasping and pushing actions in robotic manipulation, indicating efficient skill coordination.

The system achieves a completion rate of 100%, indicating effectiveness in cluttered environments.
With an average grasping success of 93.1%, the approach outperforms other policies in efficiency and performance.
The hierarchical architecture supports scalable learning by integrating reinforcement learning and behaviour cloning techniques.
Dynamic decision-making substantially enhances manipulation outcomes compared to traditional methods.

Yu et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark