Call for Papers

Scope and Topics of Interest

The aim of this workshop is to bring together researchers and practitioners in the field of policy learning for geometrical spaces. We welcome contributions on the following topics:

• Representation learning techniques for graphs, meshes, or voxels to improve reinforcement learning algorithms

• Deep metric learning for vision-based policy learning to enhance the ability of robots to operate in complex 3D environments

• Utilization of point clouds, NeRFs, neural SDFs, occupancy networks, those based on vision transformers and other advanced representations for policy learning in 3D geometric spaces to improve accuracy and efficiency

• Techniques such as occlusion maps for handling partial observability in policy learning and enabling robots to operate effectively in challenging environments

• Deep reinforcement learning algorithms for policy learning in 3D geometric spaces

• Identifying and addressing the challenges and opportunities of policy learning in 3D geometric spaces, such as sample efficiency, generalization, and stability

• Equivariant algorithms for policy and model learning

• Real-world implementation and evaluation of policy learning algorithms in 3D environments to promote the translation of research to practical applications

• Applications of policy learning in robotics, including grasping, manipulation, motion planning, and autonomous navigation, to address the growing demand for effective robot operation

Important Dates

Dates are tentative and subject to change