The Limits and Potentials of Deep Learning for Robotics
It addresses problems for robotics researchers by identifying key issues in applying deep learning, but it is incremental as it reviews existing challenges without presenting new solutions.
The paper discusses robotics-specific challenges for deep learning, such as learning, reasoning, and embodiment, and highlights the need for better evaluation metrics and the balance between data-driven and model-driven approaches to overcome current limitations.
The application of deep learning in robotics leads to very specific problems and research questions that are typically not addressed by the computer vision and machine learning communities. In this paper we discuss a number of robotics-specific learning, reasoning, and embodiment challenges for deep learning. We explain the need for better evaluation metrics, highlight the importance and unique challenges for deep robotic learning in simulation, and explore the spectrum between purely data-driven and model-driven approaches. We hope this paper provides a motivating overview of important research directions to overcome the current limitations, and help fulfill the promising potentials of deep learning in robotics.