A Deep Learning Approach to Grasping the Invisible

Keyphrases

• Deep Learning 100%
• Invisible 100%
• Target-oriented 66%
• Task Completion Rate 66%
• Large Margin 33%
• Visual Target 33%
• Visual Observation 33%
• Real Robots 33%
• Target Information 33%
• Robotic Manipulation 33%
• Robot Experiment 33%
• Action Experience 33%
• Grasping Motion 33%
• Robot-environment Interaction 33%
• Deep Q-learning 33%
• Observation Information 33%
• Emerging Problems 33%
• Motion Primitives 33%
• Future Rewards 33%
• Grasping Action 33%
• Cluttered Objects 33%

Computer Science

• Robotics 100%
• Learning Approach 100%
• Deep Learning Method 100%
• Task Success Rate 50%
• Environment Interaction 25%
• Domain Knowledge 25%
• Visual Observation 25%
• Supplementary Material 25%