Behavior trees replaced finite state machines as the default architecture for high-level robot control. Here is the formalism that makes them reactive, the BehaviorTree.CPP and ROS2 Nav2 implementation, and an honest account of where BTs genuinely beat the alternatives and where they do not.
Robotics
-
Behavior Trees for Robotics -
Humanoid Robots in 2026 Honestly A skeptical, engineering-grounded look at the 2026 humanoid-robot wave: Figure, Tesla Optimus, Unitree, 1X, Boston Dynamics Atlas, and Agility Digit. What is genuinely autonomous versus teleoperated or cherry-picked, the hard problems still unsolved in locomotion, manipulation, actuators, batteries, and VLA models, and an honest commercialization timeline.
-
Kalman Filters Explained The Kalman filter is the recursive estimator that turns noisy predictions and noisy sensors into a single best guess of where something is and how fast it is moving. We walk the predict-update loop, the actual equations, the gain as a trust dial, the EKF and UKF extensions, and the honest pain of tuning Q and R on real hardware.
-
Industrial Robotics in 2026 The industrial-robotics world in 2026 is less science-fiction than the AI headlines imply and more capable than the conventional wisdom expects. We walk what is actually on factory floors, the ISO 10218 and 15066 safety standards that shape cobot design, force-limited collaboration, the integrator economy that decides whether a robot ever ships, and the honest gap between marketing and deployment reality.
-
Inverse Kinematics for the Working Engineer Inverse kinematics is the math that turns "put the gripper here" into "set joint angles to these values." We walk forward versus inverse, analytic versus numerical solutions, the Jacobian and damped least squares, how singularities sneak up on you, the practical solvers every modern stack ships, and how to pick one without doing a PhD.
-
PID Control from First Principles PID is the feedback loop that runs an enormous fraction of industrial reality, from thermostats to flight control to chemical plants, and almost every interesting behavior of a real plant can be predicted by understanding what its three terms each do. We walk the math, the intuition, the tuning, the standard failure modes, and the honest gap between the textbook controller and the one that actually runs.
-
ROS and ROS2 Explained ROS is the Linux of robotics — not an operating system in any kernel sense, but the convention-and-message-bus stack that nearly every modern robot speaks. We walk what nodes, topics, services, and actions actually are, why ROS1's central master became ROS2's distributed DDS, the build-and-launch story, and the honest learning curve before you can move a real robot with it.
-
SLAM in Practice SLAM is the math that lets a robot build a map while figuring out where it is on the same map. We walk the chicken-and-egg structure, sensor modalities (lidar versus visual versus inertial), the particle-filter to graph-optimization shift, why loop closure matters more than any single odometry source, and the honest accuracy reality from a Roomba to a self-driving car.
-
Stepper vs Servo vs BLDC Stepper, servo, and BLDC motors look similar from the outside and behave very differently in a real project. We walk what each one actually is at the rotor level, the torque curves that decide where each wins, open-loop versus closed-loop control, field-oriented control on a BLDC, and a clear decision framework for choosing the right motor class for the job.