Blogs - RL | Generative Model | Control | Humanoid Animation

---

Survey Paper : Robot Skills & Future Directions

View my survey paper on robot skills with an interactive PDF viewer. Navigate pages, zoom in/out, and download directly from the browser without leaving the site.

---

The Bitter Lesson — Reflections on Sutton’s Talk

Notes on Richard Sutton’s “Bitter Lesson”: human heuristics vs. large-scale learning and search, and what prediction learning means for how machines (and humans) learn.

---

Flow Matching Explained

An intuitive explanation of Flow Matching for generative modeling. Learn how to directly match probability flows as an efficient alternative to diffusion models.

---

Policy Gradient & Actor-Critic Explained

Explanation of Policy Gradient and Actor-Critic methods in Reinforcement Learning, taught by Prof. Sergey Levine in CS185 2026

---

PPO: Proximal Policy Optimization

The contemporary RL algo used widely in robotics sim2real.

---

Nonlinear Systems — Complete Notes

Complete notes on nonlinear systems: stability theory, Lyapunov methods, contraction mapping, and existence/uniqueness. Taught by Prof. Koushil Sreenath. in ME 237

---

Nonlinear Systems Fun Fact

Small facts and intuitions collected while studying nonlinear systems — phase planes, $\dot{x}=f(x)$ behavior, and the geometry behind them.

---

Order of the System

Why don’t we use triple-dot or higher-order derivatives to model dynamic systems?

---

Riccati Equation

Notes on the algebraic and differential Riccati equation in optimal control (LQR).

---

Euler Angles & Gimbal Lock

Why Euler angle parameterizations of rotation suffer from gimbal lock, and what alternatives (quaternions, rotation matrices) buy you.

---

Sliding Mode Control — Reaching Speed

Notes on the reaching law and reaching speed in sliding mode control, and how they trade off chattering against convergence time.

---