Jochen Gast

• Research conducted at the intersection of computer vision and machine learning • Probabilistic vision / Variational inference methods / Deep learning • Video deblurring and optical flow methods • Teaching assistance

• Developed a framework for real-time collision detection based on continuous penalty forces. • Parallelized CPU collision detection algorithms based on CUDA and Nvidia's thrust library.

• Worked on computer vision research projects in the field of monocular depth estimation.

• A geometric multigrid method for simulating deformable models on unstructured, non-nested mesh hierarchies.

• Simulated structural mechanics via a FEM method and fluid dynamics using a FD method. • Visualized QT real-time simulations using OpenGL/OpenSG scene graph systems. • Implemented an adaptive modeling engine via lattice based free-form deformation.

• Research conducted at the intersection of computer vision and machine learning • Probabilistic vision / Variational inference methods / Deep learning • Video deblurring and optical flow methods • Teaching assistance

Computer Vision, Machine Learning, Robot Learning, Optimization of Static/Dynamic Systems, Control Theory, Fuzzy Logic, Algorithmic Modeling, Numerical Methods in Mechanics, Simulation of Deformable Bodies, Computational Fluid Dynamics. Thesis: Estimating Motion from a Single Blurry Image

Artificial Machine Intelligence, Multivariable Control, Real-time Operating Systems, Theoretical Foundations of Computing

Computer Graphics, Massively Parallel Processors (CUDA), Elementary PDE, Robotics, Engineering Mechanics, Software Engineering, Geometric Methods of CAE/CAD. Thesis: A geometric multigrid method for simulating deformable models on unstructured, non-nested mesh hierarchies