Let's Push Things Forward: A Survey on Robot Pushing

@article{Stber2020LetsPT,
  title={Let's Push Things Forward: A Survey on Robot Pushing},
  author={Jochen St{\"u}ber and Claudio Zito and R. Stolkin},
  journal={Frontiers in Robotics and AI},
  year={2020},
  volume={7}
}
As robots make their way out of factories into human environments, outer space, and beyond, they require the skill to manipulate their environment in multifarious, unforeseeable circumstances. With this regard, pushing is an essential motion primitive that dramatically extends a robot's manipulation repertoire. In this work, we review the robotic pushing literature. While focusing on work concerned with predicting the motion of pushed objects, we also cover relevant applications of pushing for… 
Deep Reinforcement Learning of the Inverse Pushing Model in Robotics
TLDR
This thesis presents a Deep Reinforcement Learning Framework approach to learn inverse models for robotic push manipulation of single objects, based on available forward pushing models, and illustrates the application of the proposed approach on a box shaped object.
Object and Relation Centric Representations for Push Effect Prediction
TLDR
This paper proposes a graph neural network based framework for effect prediction and parameter estimation of pushing actions by modeling object relations based on contacts or articulations and demonstrates 6D effect prediction in the lever-up action in the context of robot-based hard-disk disassembly.
Goal-Driven Robotic Pushing Using Tactile and Proprioceptive Feedback
TLDR
This study proposes a reactive and adaptive method for robotic pushing that uses rich feedback from a high-resolution optical tactile sensor to control push movements instead of relying on analytical or data-driven models of push interactions.
COCOI: Contact-aware Online Context Inference for Generalizable Non-planar Pushing
  • Zhuo Xu, Wenhao Yu, Daniel Ho
  • Computer Science
    2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)
  • 2021
TLDR
This work proposes Contact-aware Online COntext Inference (COCOI), a deep RL method that encodes a context embedding of dynamics properties online using contact-rich interactions and studies this method based on a novel and challenging non-planar pushing task.
Bridging the gap between learning and heuristic based pushing policies
TLDR
A heuristic rule is utilized that moves the target object towards the workspace’s empty space and achieves singulation and is incorporated to the reward in order to train more efficiently reinforcement learning agents for singulation.
Accurate Vision-based Manipulation through Contact Reasoning
TLDR
This work proposes to disentangle contact from motion optimization, and uses a hybrid approach for perception and state estimation that combines neural networks with a physically meaningful state representation for vision-based manipulation.
Learning Transferable Push Manipulation Skills in Novel Contexts
TLDR
A parametric internal model for push interactions is proposed that enables a robot to predict the outcome of a physical interaction even in novel contexts, and both biased and unbiased predictors can reliably produce predictions in line with the outcomes of a carefully tuned physics simulator.
DIPN: Deep Interaction Prediction Network with Application to Clutter Removal
TLDR
The overall network demonstrates intelligent behavior in selecting proper actions between push and grasp for completing clutter removal tasks and significantly outperforms the previous state-of-the-art.
EXI-Net: EXplicitly/Implicitly Conditioned Network for Multiple Environment Sim-to-Real Transfer
TLDR
This study proposes a network architecture with explicit and implicit dynamics parameters for sim-to-real transfer from multiple environments and applies this method to the object pushing task, verifying its effectiveness by comparing it with previous methods and real-world experiments.
Grasping and Manipulation with a Multi-Fingered Hand
TLDR
The approach presented in this report is to study and possibly extend a new approach to artificial intelligence (A.I.) which has emerged in the last years in response to the necessity of building intelligent controllers for agents operating in unstructured stochastic environments.
...
...

References

SHOWING 1-10 OF 107 REFERENCES
Pushing corridors for delivering unknown objects with a mobile robot
TLDR
This work introduces a multilayer, modular pushing skill that enables a robot to push unknown objects in such environments and proposes an adaptive pushing controller that learns local inverse models of robot-object interaction on the fly.
A Framework for Push-Grasping in Clutter
TLDR
This work introduces a framework for planning in clutter that uses a library of actions inspired by human strategies that succeeds where traditional grasp planners fail, and works under high uncertainty by utilizing the funneling effect of pushing.
Deep visual foresight for planning robot motion
  • Chelsea Finn, S. Levine
  • Computer Science
    2017 IEEE International Conference on Robotics and Automation (ICRA)
  • 2017
TLDR
This work develops a method for combining deep action-conditioned video prediction models with model-predictive control that uses entirely unlabeled training data and enables a real robot to perform nonprehensile manipulation — pushing objects — and can handle novel objects not seen during training.
Online Adaptation of Robot Pushing Control to Object Properties
  • S. Krivic, J. Piater
  • Computer Science
    2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)
  • 2018
TLDR
A data-driven approach for learning local inverse models of robot-object interaction for push manipulation with a high degree of robustness and a high success rate in pushing objects towards a fixed target and along a path compared to previous methods.
More than a million ways to be pushed. A high-fidelity experimental dataset of planar pushing
TLDR
A comprehensive and high-fidelity dataset of planar pushing experiments is presented, characterizing the variability of friction, and evaluating the most common assumptions and simplifications made by models of frictional pushing in robotics.
Push-Net: Deep Planar Pushing for Objects with Unknown Physical Properties
TLDR
This paper introduces Push-Net, a deep recurrent neural network model, which enables a robot to push objects of unknown physical properties for re-positioning and re-orientation, using only visual camera images as input, and tests suggest that it is robust and efficient.
Stable Pushing: Mechanics, Controllability, and Planning
TLDR
A planner for finding stable pushing paths among obstacles is described, and the planner is demon strated on several manipulation tasks.
Fast adaptation for effect-aware pushing
TLDR
A mathematical compact model for planar sliding motion of an object is built and it is shown how a robot acquires the parameters of such a model; then how this is used to predict pushing actions and to move an object from any1 position and orientation to another.
Omnipush: accurate, diverse, real-world dataset of pushing dynamics with RGB-D video
TLDR
Omnipush is introduced, a dataset with high variety of planar pushing behavior, constructed so as to systematically explore key factors that affect pushing–the shape of the object and its mass distribution–which have not been broadly explored in previous datasets, and allow to study generalization in model learning.
A robust pushing skill for object delivery between obstacles
TLDR
The concept of a pushing corridor for cluttered environments that leaves the robot sufficient space for corrective motions is introduced and a reactive manipulation skill for pushing objects along this collision-free corridor is proposed.
...
...