Sparse summarization of robotic grasping data

We propose a new approach for learning a summarized representation of high dimensional continuous data. Our technique consists of a Bayesian non-parametric model capable of encoding high-dimensional data from complex distributions using a sparse summarization. Specifically, the method marries techniques from probabilistic dimensionality reduction and clustering. We apply the model to learn efficient representations of grasping data for two robotic scenarios.

Authors

• Martin Hjelm
• Carl Henrik Ek
• Renaud Detry
• Hedvig Kjellström
• Danica Kragic

Keywords

• Robots
• Data models
• Grasping
• Shape
• Kernel
• Optimization
• Encoding
• Data Summarization
• Robotic Grasping
• Dimensionality Reduction
• High Diversity
• Latent Variables
• Similarity Measure
• Local Minima
• Data Augmentation
• Energy Function
• Data Clustering
• Kernel Function
• Latent Space
• Discrete Data
• Gaussian Process
• Cluster Centers
• Number Of Centers
• Spectral Method
• Covariance Function
• Random Initialization
• Latent Representation
• Latent Variable Model
• Cluster Representatives