Applications of Deep Learning in Robot Vision

摘要

Deep learning techniques developed in robot soccer can therefore be transferred to other resource-constrained platforms, such as smartphones and Unmanned Aerial Vehicles. Deep learning is a hot topic in the pattern recognition, machine learning, and computer vision research communities. Deep learning has already attracted the attention of the robot vision community. The chapter is intended to be a guide for developers of robot vision systems and discusses requirements for robot vision applications. It describes relevant papers related to robot vision and also discusses current and future challenges of deep learning in robotic applications. The chapter explores advances based on deep learning related to robot vision and future challenges. Unsupervised learning is another relevant area of future development for deep learning-based vision systems. Biological learning is largely unsupervised; animals and humans discover the structure of the world by exploring and observing it.