Project Icon

theseus

Optimize Robotics and Vision with Differentiable Nonlinear Library

Product DescriptionTheseus is a versatile PyTorch library for nonlinear optimization layers in robotics and vision, featuring differentiable optimization for enhancing neural models. It includes user-centric tools for various optimizers such as Gauss-Newton and Levenberg-Marquardt, supports dense and sparse linear solvers, and has functionalities for robotic kinematics and lie groups. Designed for computational efficiency, it offers GPU acceleration and seamless integration via PyPI or source.
Project Details