Jekyll2026-10-02T08:36:16+07:00https://nhattran-97.github.io/feed.xmlNhat TranPersonal portfolio of Nhat Tran — robotics & AI research, projects, publications and technical notes.Kalman Filter — A Practical Cheat Sheet2026-09-20T00:00:00+07:002026-09-20T00:00:00+07:00https://nhattran-97.github.io/notes/2026/kalman-filter-cheatsheetA short summary of the linear Kalman filter: the model, the two steps, and the tuning tips I always forget.

System model

\[x_k = F x_{k-1} + B u_k + w_k, \qquad w_k \sim \mathcal{N}(0, Q)\] \[z_k = H x_k + v_k, \qquad v_k \sim \mathcal{N}(0, R)\]

1. Predict

\[\hat{x}_{k|k-1} = F \hat{x}_{k-1} + B u_k, \qquad P_{k|k-1} = F P_{k-1} F^\top + Q\]

2. Update

\[K_k = P_{k|k-1} H^\top \left(H P_{k|k-1} H^\top + R\right)^{-1}\] \[\hat{x}_k = \hat{x}_{k|k-1} + K_k \left(z_k - H \hat{x}_{k|k-1}\right), \qquad P_k = (I - K_k H) P_{k|k-1}\]

Minimal Python implementation

import numpy as np

class KalmanFilter:
    def __init__(self, F, H, Q, R, x0, P0):
        self.F, self.H, self.Q, self.R = F, H, Q, R
        self.x, self.P = x0, P0

    def predict(self):
        self.x = self.F @ self.x
        self.P = self.F @ self.P @ self.F.T + self.Q

    def update(self, z):
        S = self.H @ self.P @ self.H.T + self.R
        K = self.P @ self.H.T @ np.linalg.inv(S)
        self.x = self.x + K @ (z - self.H @ self.x)
        self.P = (np.eye(len(self.x)) - K @ self.H) @ self.P

Tuning tips

Symptom Likely cause Fix
Estimate lags behind measurements $Q$ too small Increase $Q$
Estimate is noisy $R$ too small Increase $R$
Filter diverges Model mismatch / bad $P_0$ Check $F$, inflate $P_0$

Takeaway: the ratio between $Q$ and $R$ matters more than their absolute values.

]]>ROS 2 Commands I Use Every Day2026-08-10T00:00:00+07:002026-08-10T00:00:00+07:00https://nhattran-97.github.io/notes/2026/ros2-cheatsheetA living list of ROS 2 CLI commands for debugging nodes, topics and TF.

Workspace

colcon build --symlink-install --packages-select my_pkg
source install/setup.bash

Topics

ros2 topic list -t                 # list with message types
ros2 topic echo /odom --once       # print a single message
ros2 topic hz /scan                # check publish rate

TF

ros2 run tf2_ros tf2_echo map base_link
ros2 run tf2_tools view_frames      # generates frames.pdf

Notes

  • Use --symlink-install so Python and launch file edits take effect without rebuilding.
  • ros2 doctor --report is a quick sanity check when something feels off.
]]>How I Read a Research Paper2026-07-02T00:00:00+07:002026-07-02T00:00:00+07:00https://nhattran-97.github.io/notes/2026/how-to-read-a-paperMy three-pass method for reading papers efficiently, adapted from S. Keshav.

Pass 1 — The bird’s-eye view (5–10 min)

Read the title, abstract, introduction, section headings and conclusion. Decide: is this paper worth reading further?

Pass 2 — The content (~1 hour)

Look carefully at figures and tables. Note the key claims and the evidence for each. Mark unfamiliar references to read later.

Pass 3 — Virtual re-implementation (several hours)

Try to reproduce the paper in your head (or in code). This is where you find hidden assumptions and weaknesses.

Template I keep for every paper

**Problem:**
**Key idea:**
**Method:**
**Results:**
**Limitations:**
**What I can reuse:**
]]>