Learning a visuomotor controller for real world robotic grasping using simulated depth images

By Ulrich Viereck, Andreas ten Pas, Kate Saenko, Robert W. Platt. 2017

197 citations recorded by the source corpus.

We want to build robots that are useful in unstructured real world applications, such as doing work in the household. Grasping in particular is an important skill in this domain, yet it remains a challenge. One of the key hurdles is handling unexpected changes or motion in the objects being grasped and kinematic noise or other errors in the robot. This paper proposes an approach to learning a closed-loop controller for robotic grasping that dynamically guides the gripper to the object. We use a wrist-mounted sensor to acquire depth images in front of the gripper and train a convolutional neural network to learn a distance function to true grasps for grasp configurations over an image. The training sensor data is generated in simulation, a major advantage over previous work that uses real robot experience, which is costly to obtain. Despite being trained in simulation, our approach works well on real noisy sensor images. We compare our controller in simulated and real robot experiments to a strong baseline for grasp pose detection, and find that our approach significantly outperforms the baseline in the presence of kinematic noise, perceptual errors and disturbances of the object during grasping.

Subfields: Biomedical Engineering, Control and Systems Engineering

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

In Control and Systems Engineering

Citations in the corpus

Driving Policy Transfer via Modularity and Abstraction (2018, background): “to facilitate transfer: for example, variants of domain adaptation [48,16,51,4,36] and specialized architectures [41,53]. Working with depth maps instead of color images is known to simplify transfer [50,28,27]. When dealing with color images, domain randomization [46,21,42,43] enables direct sim-to”
QT-Opt: Scalable Deep Reinforcement Learning for Vision-Based Robotic Manipulation (2018, background): “While most prior grasping methods operate in open-loop, a number of works have studied closed-loop grasping [24, 25, 26, 27].”
Asymmetric Actor Critic for Image-Based Robot Learning (2017, background): “However, [11], [17], [19] show how randomizing textures and lighting allows for effective transfer.”
Modular Deep Q Networks for Sim-to-real Transfer of Visuo-motor Policies (2016, unknown): “Recently it has been proposed to simulate depth images to learn and then transfer grasping skills to real-world robotic arms [Viereck et al. , 2017], yet no adaptation in the real-world has been performed.”
Closing the Loop for Robotic Grasping: A Real-time, Generative Grasp Synthesis Approach (2018, methodology): “Thirdly, we show our system’s ability to generalise to dynamic cluttered scenes by reproducing the experiments from [32] and show improved results.”
RT-1: Robotics Transformer for Real-World Control at Scale (2022, background): “A significant body of work deals with learning policies and predictive models for robotic grasping [53, 36, 46, 14, 64], with the aim of generalizing to new objects.”
Grasping in the Wild: Learning 6DoF Closed-Loop Grasping From Low-Cost Demonstrations (2019, background): “However, most prior self-supervised grasping systems are often constrained to learning in simulation [15], [17], [20], [24] or structured lab environments [1], [3], [21], [22].”
The Feeling of Success: Does Touch Sensing Help Predict Grasp Outcomes? (2017, unknown): “As a substitute for this capability, some works have used wrist-mounted depth cameras to obtain a better estimate of local geometry [20].”
Sim-to-real Transfer of Visuo-motor Policies for Reaching in Clutter: Domain Randomization and Adaptation with Modular Networks (2017, background): “Recently it has been proposed to simulated depth images to learn and then transfer grasping skills to real-world robotic arms [17], yet no adaptation in the real-world has been performed.”
Learning robust, real-time, reactive robotic grasping (2019, methodology): “Despite our GG-CNN not being trained on cluttered scenes, we achieved a grasp success rate of 87% (83/96) compared with 89% (66/74) in Viereck et al. (2017).”
Learning Deep Policies for Robot Bin Picking by Simulating Robust Grasping Sequences (2017, methodology): “[11] used CNNs trained to predict antipodal grasps on dense 3D point clouds of a scene from synthetic datasets.”
How to train your robot with deep reinforcement learning: lessons we have learned (2021, background): “A number of works have studied closed loop grasping (Yu and Rodriguez 2018; Viereck et al. 2017; Hausman et al. 2017; Levine et al. 2018).”