Abstract:Human dexterity is an invaluable capability for precise manipulation of objects in complex tasks. The capability of robots to similarly grasp and perform in-hand manipulation of objects is critical for their use in the ever changing human environment, and for their ability to replace manpower. In recent decades, significant effort has been put in order to enable in-hand manipulation capabilities to robotic systems. Initial robotic manipulators followed carefully programmed paths, while later attempts provided a solution based on analytical modeling of motion and contact. However, these have failed to provide practical solutions due to inability to cope with complex environments and uncertainties. Therefore, the effort has shifted to learning-based approaches where data is collected from the real world or through a simulation, during repeated attempts to complete various tasks. The vast majority of learning approaches focused on learning data-based models that describe the system to some extent or Reinforcement Learning (RL). RL, in particular, has seen growing interest due to the remarkable ability to generate solutions to problems with minimal human guidance. In this survey paper, we track the developments of learning approaches for in-hand manipulations and, explore the challenges and opportunities. This survey is designed both as an introduction for novices in the field with a glossary of terms as well as a guide of novel advances for advanced practitioners.

What problem does this paper attempt to address?

The problem this paper attempts to address is the lack of in-hand manipulation capabilities in robots. Specifically, the paper aims to review and analyze the application and development of learning-based methods in the field of in-hand manipulation, exploring how these methods enable robots to dexterously grasp and manipulate objects like humans, especially in complex and uncertain environments. The paper particularly focuses on three main learning methods: model-based learning, reinforcement learning (RL), and imitation learning (IL), and discusses their advantages and challenges in improving the in-hand manipulation capabilities of robots. The paper points out that traditional in-hand manipulation methods, such as precise programming-based path following or motion and contact-based analytical modeling, have failed to provide practical solutions due to their inability to cope with complex environments and uncertainties. Therefore, the research focus has shifted to learning-based methods, which improve the adaptability and flexibility of robots by collecting data from real-world or simulated environments and training robots through repeated attempts to complete various tasks. Reinforcement learning, in particular, has garnered increasing attention for its ability to generate problem-solving strategies with minimal human guidance. In summary, this paper aims to provide a comprehensive guide through a thorough analysis of existing learning methods, helping both novice and advanced researchers better understand and advance the development of in-hand manipulation technology in robots.

Survey of Learning-based Approaches for Robotic In-Hand Manipulation

Survey of Learning Approaches for Robotic In-Hand Manipulation

A Survey of Embodied Learning for Object-Centric Robotic Manipulation

Robotic in-hand manipulation with relaxed optimization

A survey of robot manipulation in contact

A Survey on Deep Reinforcement Learning Algorithms for Robotic Manipulation

A Survey of Multifingered Robotic Manipulation: Biological Results, Structural Evolvements, and Learning Methods

A Survey on Learning-Based Robotic Grasping

Learning by Watching: A Review of Video-based Learning Approaches for Robot Manipulation

Review of Learning-Based Robotic Manipulation in Cluttered Environments

Machine Learning Meets Advanced Robotic Manipulation

Tactile Active Inference Reinforcement Learning for Efficient Robotic Manipulation Skill Acquisition

Reinforcement Learning for Robotic Grasping and Manipulation: A Review

Survey of Imitation Learning for Robotic Manipulation

Learning Synergies Based In-Hand Manipulation with Reward Shaping

Dexterous Manipulation with Deep Reinforcement Learning: Efficient, General, and Low-Cost

Learning for a Robot: Deep Reinforcement Learning, Imitation Learning, Transfer Learning

A Review on Robot Manipulation Methods in Human-Robot Interactions

Learning Hierarchical Control for Robust In-Hand Manipulation