Tags / 7 articles

control

Every article tagged control, newest first.

Precise control of the SO-101 arm · Part 7 of 8October 15, 202612 min read

Two bar charts of the same Measured N64 comparison. At fixed time, baseline error is 2.238 mm and frozen-head error is 1.643 mm. With lag fitted using native tablet reports, baseline error is 1.537 mm and frozen-head error is 1.641 mm.

Teaching my drawing arm to keep learning

After part 6: direct motor drive, learning between drawings, feedback during a stroke, and a twin tested across days. The repeat floor is not a claim of one-millimeter control.

Precise control of the SO-101 arm · Part 4 of 8October 5, 202622 min read

Number line in ticks: reading 100, next reading 110, goal sent 150. The joint moved 10 and the label, the goal minus the next reading, is 40.

Learning what the physics misses

Where a neural network fits in a robot controller. A network that learns only what the physics model gets wrong does the same job as a Kalman disturbance estimate, with more freedom, and this is where recent research is going.

so-101 control machine-learning

Precise control of the SO-101 arm · Part 3 of 8October 4, 202635 min read

Bar chart of the real-arm tracking error. Cat motion: direct 22.5 mrad at 30 Hz, pi 8.4, solve 6.2 and mpc 6.0 at 60 Hz. Signature motion: direct 28.3, lead 23.7 and inv 13.4 mrad.

Six ways to choose the goal

One general formula for the goal sent to a servo. Every controller I built, from a plain look-ahead to MPC, is that formula with some terms set to zero.

so-101 control

Precise control of the SO-101 arm · Part 2 of 8October 3, 202616 min read

Top: a ramp of true angles and the staircase of readings rounded to the nearest tick. Bottom: the reading error, a sawtooth between minus and plus half a tick, with an RMS of 1 over the square root of 12 tick, 0.443 mrad.

Seeing finer than the sensor

The sensor side of the joint equation: why differences of encoder ticks explode the noise, and how an observer or a Kalman filter can estimate an angle finer than one tick.

so-101 control sensors

Precise control of the SO-101 arm · Part 1 of 8October 2, 202644 min read

Three drawings of one joint as a torsion spring. With no load, the goal and the joint sit on the target. With gravity and the goal on the target, the joint sags below and the spring between goal and joint stretches. With the goal moved up past the target, the stretched spring holds the joint on the target.

One equation, and every way the arm misses it

Start from torque = inertia × acceleration, then add each term a cheap servo arm really has: gravity, friction, a payload, dead time, a dead band. Each term is one way the arm misses its target.

so-101 control hardware

September 30, 202627 min read

Before a cube moves, the robot gripper surrounds it. After it moves, the same grasp closes at the old position.

How robot control is changing

From copying a person to practicing and imagining: how a robot chooses its next movement in 2026, and what is still missing.

control reinforcement-learning industry

September 26, 202617 min read

Franka arm with human-readable joint names and colored rotation arrows

How an arm moves: joints, geometry, and motor control

From encoder readings to gripper position: a visual guide to robot geometry, servos, and the commands that connect them.

control hardware sensors

← All tags · All writing