Grasp2Vec: Learning Object Representations from Self-Supervised Grasping

By Eric Jang, Coline Devin, Vincent Vanhoucke, Sergey Levine. 2018

119 citations recorded by the source corpus.

Well structured visual representations can make robot learning faster and can improve generalization. In this paper, we study how we can acquire effective object-centric representations for robotic manipulation tasks without human labeling by using autonomous robot interaction with the environment. Such representation learning methods can benefit from continuous refinement of the representation as the robot collects more experience, allowing them to scale effectively without human intervention. Our representation learning approach is based on object persistence: when a robot removes an object from a scene, the representation of that scene should change according to the features of the object that was removed. We formulate an arithmetic relationship between feature vectors from this observation, and use it to learn a representation of scenes and objects that can then be used to identify object instances, localize them in the scene, and perform goal-directed grasping tasks where the robot must retrieve commanded objects from a bin. The same grasping procedure can also be used to automatically collect training data for our method, by recording images of scenes, grasping and removing an object, and recording the outcome. Our experiments demonstrate that this self-supervised approach for tasked grasping substantially outperforms direct reinforcement learning from images and prior representation learning methods.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Control and Systems Engineering, Reinforcement Learning, Representation & Self-Supervised Learning

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Control and Systems Engineering

In Reinforcement Learning

In Representation & Self-Supervised Learning

Citations in the corpus

S4L: Self-Supervised Semi-Supervised Learning (2019, background): “Self-supervised techniques have a variety of applications in a broad range of computer vision topics [14, 35, 7, 29, 34].”
Revisiting Self-Supervised Visual Representation Learning (2019, background): “com/google/revisiting-self-supervised In robotics, both the result of interacting with the world, and the fact that multiple perception modalities simultaneously get sensory inputs are strong signals which can be exploited to create self-supervised tasks [21, 41, 26, 10].”
Self-Supervised GANs via Auxiliary Rotation Loss (2018, methodology): “It has been widely used in the video domain [39, 40], the robotics domain [41, 42] and the image domain [24, 35].”
High-Fidelity Image Generation With Fewer Labels (2019, methodology): “This approach was successfully applied to different data modalities, such as images (Doersch et al., 2015; Caron et al., 2018), video (Agrawal et al., 2015; Lee et al., 2017), and robotics (Jang et al., 2018; Pinto & Gupta, 2016).”
Charting the Right Manifold: Manifold Mixup for Few-shot Learning (2019, background): “Self-supervision techniques have been successfully applied to diverse set of domains, ranging from robotics to computer vision [31, 12, 57, 55, 45].”
A Review of Robot Learning for Manipulation: Challenges, Representations, and Algorithms (2019, background): “Neural networks are also highly effective at combining data from multiple sensor modalities or information sources [119, 233, 236, 370, 369, 240, 173].”
Learning Rope Manipulation Policies Using Dense Object Descriptors Trained on Synthetic Depth Data (2020, background): “Manipulating deformable objects is valuable for a wide variety of applications from surgery and manufacturing to household robotics [2, 8, 11, 14, 15, 19, 26, 37–39].”
Bisimulation Makes Analogies in Goal-Conditioned Reinforcement Learning (2022, methodology): “We take inspiration from works that have shown the arithmetic properties of embeddings (Mikolov et al., 2013; Jang et al., 2018) to enforce the same type of intuitive structure in the state encoder: From Deﬁnition 4.1 we know that if there exists s j ∈ S , g j ∈ G such that φ ( s i , g i ) = φ ( s…”
Self-Supervised Goal-Conditioned Pick and Place (2020, background): “Several prior approaches have shown that robotic grasping data can be used to learn object embeddings useful for goal-conditioned grasping [5] and assembly [12].”
Does Unsupervised Architecture Representation Learning Help Neural Architecture Search? (2020, unknown): “Unsupervised representation learning has been successfully used in a wide range of domains including natural language processing [1, 2, 3], computer vision [4, 5], robotic learning [6, 7], and network analysis [8, 9].”