LaSeSOM: A Latent and Semantic Representation Framework for Soft Object Manipulation

Zhou, Peng; Zhu, Jihong; Huo, Shengzeng; Navarro-Alarcon, David

doi:10.48550/arxiv.2012.05412

Cited by 1 publication

(1 citation statement)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DAE is a type of artificial neural network with three or more layers trained in the unsupervised-learning manner. It can automatically learn features from unlabeled data and give a better description than the original data [32]. The traditional DAE [8] consists of an Encoder and Decoder, shown in Fig.…”

Section: Feature Extractionmentioning

confidence: 99%

Towards Latent Space Based Manipulation of Elastic Rods using Autoencoder Models and Robust Centerline Extractions

Qi¹,

Ma²,

Zhou³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

The automatic shape control of deformable objects is an open (and currently hot) manipulation problem that is challenging due to the object's high-dimensional shape information and its complex physical properties. As a feasible solution to these issues, in this paper, we propose a new methodology to automatically deform elastic rods into 2D desired shapes. For that, we present an efficient vision-based controller that uses a deep autoencoder network to compute a compact representation of the object's infinite dimensional shape. To deal with the (typically unknown) mechanical properties of the object, we use an online algorithm that approximates the sensorimotor mapping between the robot's configuration and the object's shape features. Our new approach has the capability to compute the rod's centerline from raw visual data in real-time; This is done by introducing an adaptive algorithm based on a self-organizing network. The effectiveness of the proposed method is thoroughly validated with simulations and experiments.

show abstract

Section: Feature Extractionmentioning

confidence: 99%