Autonomous Curiosity for Real-Time Training Onboard Robotic Agents

Abstract

Learning requires both study and curiosity. A good learner is not only good at extracting information from the data given to it, but also skilled at finding the right new information to learn from. This is especially true when a human operator is required to provide the ground truth - such a source should only be queried sparingly. In this work, we address the problem of curiosity as it relates to online, real-time, human-in-the-loop training of an object detection algorithm onboard a robotic platform, one where motion produces new views of the subject. We propose a deep reinforcement learning approach that decides when to ask the human user for ground truth, and when to move. Through a series of experiments, we demonstrate that our agent learns a movement and request policy that is at least 3x more effective at using human user interactions to train an object detector than untrained approaches, and is generalizable to a variety of subjects and environments.

arXiv comment10 pages, 9 figures. Accepted in IEEE Winter Conference on Applications of Computer Vision (WACV), 2019. arXiv admin note: text overlap with arXiv:1902.01569

References in corpus (12)

• SSD: Single Shot MultiBox Detector
• Dueling Network Architectures for Deep Reinforcement Learning
• Deep Recurrent Q-Learning for Partially Observable MDPs
• Cost-Effective Active Learning for Deep Image Classification
• Unity: A General Platform for Intelligent Agents
• Control of a Quadrotor with Reinforcement Learning
• Learning to Navigate in Complex Environments
• Deep Interactive Object Selection
• On Learning Where To Look
• Training object class detectors with click supervision
• Click Carving: Segmenting Objects in Video with Point Clicks
• End-to-end Active Object Tracking and Its Real-world Deployment via Reinforcement Learning