A General Approach to State Refinement

Deep learning algorithms such as Convolutional Neural Networks (CNNs) are currently used to solve a range of robotics and computer vision problems. These networks typically estimate the desired representation in a single forward pass and must therefore learn to converge from a wide range of initial conditions to a precise result. This is challenging, and has led to increased interest in the development of separate refinement modules which learn to improve a given initial estimate, thus reducing the required search space. Such modules are usually developed ad-hoc for each given application, often requiring significant engineering investment. In this work we propose a generic innovation-based CNN. Our CNN is implemented along with a stochastic gradient descent (SGD) algorithm to iteratively refine a given initial estimate. The proposed approach provides a general framework for the development of refinement modules applicable to a wide range of robotics problems. We apply this framework to object pose estimation and depth estimation and demonstrate significant improvement over the initial estimates, in the range of 4. 2 - 8. 1%, for both applications.

Authors

• Gerard Kennedy
• Jin Gao
• Zheyu Zhuang
• Xin Yu 0002
• Robert E. Mahony

Keywords

• Deep learning
• Computer vision
• Pose estimation
• Network architecture
• Convolutional neural networks
• Task analysis
• Intelligent robots
• Convolutional Neural Network
• Visual Impairment
• Stochastic Gradient Descent
• Range Of Problems
• Depth Estimation
• Iterative Refinement
• Computer Vision Problems
• Human Pose Estimation
• Object Pose
• Loss Function
• Root Mean Square Error
• Step Size
• Single Image
• Recent Approaches
• Bounding Box
• Image Pairs
• Vector Field
• RGB Images
• Fine Details
• Baseline Network
• Depth Map
• Object Depth
• Center Of Rotation
• Camera Pose
• Refinement Network
• Standard Metrics
• Refinement Step
• State Estimation Algorithm
• General Area