Recent advances in Deep Learning (DL) have led to a number of spectacular applications, ranging from self-driving cars, and (autonomous) robots that outperform humans in various tasks, to building algorithms that can understand and answer questions by fusing information from many modalities. By integrating the ability of DL and deep reinforcement learning (DRL) to learn complex behaviours, one can efficiently train robots that can interact with all the objects around them.
However, the application of DL in robotics is not trivial and leads to specific learning, reasoning and embodiment problems and research questions that are typically not addressed by the computer vision and machine learning communities. Therefore, DL itself is an important area to be investigated, rather than only focussing on its applications.
The first objective pursued through this workshop is to show and present how using artificial intelligence (AI) and DL in robotics can highly improve people’s lives. In addition, after the pandemic situation that has unfolded worldwide over the past few months, with uncertainties and the need to try to maximize all the resources in different sectors, especially in healthcare, it is crucial to raise awareness of the multiple and flexible solutions that robotics can provide. The second objective of the workshop is to show the importance of having an open DL toolkit for the robotics community, a fact that has been identified as an important goal of the European Union (EU) research in robotics.
While this workshop will focus on the design of an open toolkit for DL, the discussions will not be restricted to DL, but we will initiate fruitful discussions to have further open toolkits for various robotic methods.
DL models are becoming more and more complex, requiring vast amounts of computation power and energy. These requirements are becoming especially limiting in many robotics applications, where significant energy and computation power constraints exist.
In summary, the need for open toolkits in robotics, such as a DL toolkit that contains easy to train and deploy, real-time, lightweight, ROS-compliant DL models for robotics, is a must. This workshop aims to gather experts to share their experiences in terms of opportunities and challenges in using further open toolkits in robotics.
We believe the more researchers with access to open tools and data required to develop AI-based systems, the more innovations we will have.
Deep Learning for Robotics
Deep robot active perception and cognition
Deep robot decision making
If you haven't registered yet to attend IROS, click in the button below.