Towards Vision-Based Deep Reinforcement Learning for Robotic Motion Control

Abstract

This paper introduces a machine learning based system for controlling a robotic manipulator with visual perception only. The capability to autonomously learn robot controllers solely from raw-pixel images and without any prior knowledge of configuration is shown for the first time. We build upon the success of recent deep reinforcement learning and develop a system for learning target reaching with a three-joint robot manipulator using external visual observation. A Deep Q Network (DQN) was demonstrated to perform target reaching after training in simulation. Transferring the network to real hardware and real observation in a naive approach failed, but experiments show that the network works when replacing camera images with synthetic images.

8 pages, to appear in the proceedings of Australasian Conference on Robotics and Automation (ACRA) 2015

References in corpus (1)

• Learning Contact-Rich Manipulation Skills with Guided Policy Search

Cited by in corpus (45)

• Robust Physical-World Attacks on Deep Learning Models
• Deep Reinforcement Learning Attitude Control of Fixed-Wing UAVs Using Proximal Policy Optimization
• Transferring End-to-End Visuomotor Control from Simulation to Real World for a Multi-Stage Task
• A Review of Robot Learning for Manipulation: Challenges, Representations, and Algorithms
• Real-Time Deep Learning Approach to Visual Servo Control and Grasp Detection for Autonomous Robotic Manipulation
• Towards Cognitive Exploration through Deep Reinforcement Learning for Mobile Robots
• Model-Predictive Policy Learning with Uncertainty Regularization for Driving in Dense Traffic
• A Survey of Deep Network Solutions for Learning Control in Robotics: From Reinforcement to Imitation
• 3D Simulation for Robot Arm Control with Deep Q-Learning
• Deep Reinforcement Learning for Robotic Manipulation-The state of the art
• One-Shot Reinforcement Learning for Robot Navigation with Interactive Replay
• The Best of Both Modes: Separately Leveraging RGB and Depth for Unseen Object Instance Segmentation
• Adaptive Optics control using Model-Based Reinforcement Learning
• Towards a Common Implementation of Reinforcement Learning for Multiple Robotic Tasks
• Learning Deployable Navigation Policies at Kilometer Scale from a Single Traversal
• Differentially Private Data Generative Models
• Visual Servoing from Deep Neural Networks
• DeepHunter: Hunting Deep Neural Network Defects via Coverage-Guided Fuzzing
• Imitation Learning with Additional Constraints on Motion Style using Parametric Bias
• Unseen Object Instance Segmentation for Robotic Environments
• Adversarial Discriminative Sim-to-real Transfer of Visuo-motor Policies
• Affordance-based Reinforcement Learning for Urban Driving
• TWIMP: Two-Wheel Inverted Musculoskeletal Pendulum as a Learning Control Platform in the Real World with Environmental Physical Contact
• OffWorld Gym: open-access physical robotics environment for real-world reinforcement learning benchmark and research
• Cross-Domain Perceptual Reward Functions
• CAGFuzz: Coverage-Guided Adversarial Generative Fuzzing Testing of Deep Learning Systems
• Sample-Efficient Training of Robotic Guide Using Human Path Prediction Network
• Environmentally Adaptive Control Including Variance Minimization Using Stochastic Predictive Network with Parametric Bias: Application to Mobile Robots
• Incremental Reinforcement Learning --- a New Continuous Reinforcement Learning Frame Based on Stochastic Differential Equation methods
• Team NCTU: Toward AI-Driving for Autonomous Surface Vehicles -- From Duckietown to RobotX
• FIXAR: A Fixed-Point Deep Reinforcement Learning Platform with Quantization-Aware Training and Adaptive Parallelism
• Understanding Convolutional Networks with APPLE : Automatic Patch Pattern Labeling for Explanation
• Ultrasound-Guided Robotic Navigation with Deep Reinforcement Learning
• Taking Care of The Discretization Problem: A Comprehensive Study of the Discretization Problem and A Black-Box Adversarial Attack in Discrete Integer Domain
• DRL: Deep Reinforcement Learning for Intelligent Robot Control -- Concept, Literature, and Future
• Hybrid analysis and modeling, eclecticism, and multifidelity computing toward digital twin revolution
• Orthogonal Policy Gradient and Autonomous Driving Application
• Training an Interactive Humanoid Robot Using Multimodal Deep Reinforcement Learning
• Tuning Modular Networks with Weighted Losses for Hand-Eye Coordination
• Evaluating task-agnostic exploration for fixed-batch learning of arbitrary future tasks
• Practical Robot Learning from Demonstrations using Deep End-to-End Training
• When Autonomous Systems Meet Accuracy and Transferability through AI: A Survey
• Towards a Sample Efficient Reinforcement Learning Pipeline for Vision Based Robotics
• Transferring Agent Behaviors from Videos via Motion GANs
• Defenses Against Multi-Sticker Physical Domain Attacks on Classifiers