Aug 2026· International Journal for Research in Applied Science and Engineering Technology· Vol 14, pp. 907-913· 0 citations
TL;DR
A survey examines where reinforcement learning algorithms currently stand within this field, and lays out the core theory and terminology involved, and highlights the principal obstacles that still restrict these algorithms from being deployed on real-world robotic problems.
Abstract
Neural-network-based reinforcement learning has driven a broad set of breakthroughs across many application areas.
In the specific case of robotic manipulation, these methods raise the prospect of machines acquiring dexterous, human-like
manipulation skills straight from visual input. This survey examines where reinforcement learning algorithms currently stand
within this field. It lays out the core theory and terminology involved, and highlights the principal obstacles that still restrict
these algorithms from being deployed on real-world robotic problems. The paper closes with the authors' outlook on promising
directions for continued research.
A Movement Primitive learning algorithm based on Gaussian Processes, combined with real-time zero-shot adaptation through Pathwise Conditioning, which encapsulates the predictive uncertainty of the demonstrated movement using heteroscedastic GPs and utilizes an update via Matheron's rule to instantaneously adjust the t...
Adrián Prados, L. Lishan, Alberto Mendez et al.· Jornadas de Automática· 0 citations
Over the course of a lifetime, robots may encounter novel scenarios unaccounted for in its original training that result in performance degradation. One common approach to mitigating this issue is to further grow the offline training dataset in hopes of producing a policy robust to these changes. In contrast, biologica...
Teeratham Vitchutripop, Alyssa Quarles, Wen-He Zhang et al.· 0 citations
A unified framework that combines centralized training with decentralized execution (CTDE) and a Hybrid Reward Architecture (HRA) is introduced that enables multiple actors to share a centralized multi-head critic and substantially improves both sample efficiency and policy performance.
Changhao Li, Yifang Zhang, Heng Zhang et al.· 0 citations
Although reinforcement learning (RL) has shown great potential in legged robot motion control, traditional methods face two major bottlenecks: the strategy training of a single robot requires a large amount of computational resources, and the trained models are highly specific and complex for direct transfer and applic...
Hao-Dong Huang, Shi-Long Sun, Hailin Huang et al.· Journal of Physics, Conferen...· 1 citation· ⚡1
The rapid progress of artificial intelligence is reshaping robotics and accelerating the adoption of learning-based approaches. While purely data-driven methods have achieved remarkable success in computer vision and natural language processing, robotics remains constrained by limited data, complex real-world interacti...
Mattia Piccinini, L. Schulze, Alice Plebe et al.· 0 citations
