Model-Based Control and Learned Policies
A short comparison of model-based control and learning-based policies, separating what is well established from interpretation and open questions.
advanced computing & robotics
Insights
Short perspectives on advanced computing and robotics: concept explanations, comparisons of technical approaches, methodological notes, and open questions. Each one keeps three things apart: what is established, how it is interpreted, and what remains open.
When Is a Kalman Filter Enough?
A short comparison of the Kalman filter, its nonlinear variants and particle filters, with a rough guide to choosing between them and a list of what stays hard.
Is the Sim-to-Real Gap Something to Close or to Measure?
A short research question on the gap between simulated and real robot behaviour for learned controllers, and why it may be more useful as a measured property than as a defect to remove.
Read insight: Is the Sim-to-Real Gap Something to Close or to Measure?
What Makes a Simulation Trustworthy?
A short methodological note on verification, validation, and the limits of what a simulation can show.
Floating-Point Rounding in Long Simulations
A short methodological note on how finite precision, summation order and accumulated rounding error can matter in simulations that run for many steps.
Compute-Bound or Memory-Bound? The Roofline Idea
A short concept explanation of the roofline idea: how arithmetic throughput and memory bandwidth together bound performance, and what that implies for choosing what to optimise.
Read insight: Compute-Bound or Memory-Bound? The Roofline Idea