Christopher Towell

Papers

Total Citations

H-Index

About

Christopher Towell is a researcher whose work sits at the intersection of machine learning, robotics, and control theory, with a particular focus on value function approximation for reinforcement learning. His key contributions center on developing novel mathematical frameworks that allow robots to learn motor control more efficiently by leveraging the geometry of non-linear manifolds. In his most cited work, "Geodesic Gaussian kernels for value function approximation" (2008, 36 citations), Towell introduced a method that adapts the popular Gaussian kernel to respect the underlying structure of curved, non-linear state spaces, enabling smoother and more accurate approximations. This built on his earlier foundational paper, "Value Function Approximation on Non-Linear Manifolds for Robot Motor Control" (2007, 15 citations), which demonstrated how least-squares approaches could be made effective even when dealing with the discontinuities that naturally arise in real-world robotic tasks. Towell’s work is notable for bridging theoretical rigor with practical robotics, offering tools that help machines learn complex, continuous motor skills from limited data. His research continues to influence how reinforcement learning is applied to physical systems, making him a key figure in the development of more adaptive and intelligent robots.