PYTHON · NUMPY

The body decides what the joules reveal

Show that the compute-to-actuation ratio is set by the workload, not the chassis, and that it inverts the common assumption.

01Challenge

You are told a robot is 'mostly motors, with a computer bolted on', so hidden computation will always be lost in the actuation noise. Test that on measured data from a real autonomous mobile robot, then decide whether the claim survives.

Predict firstA mobile robot navigating autonomously. What share of its electrical power do the motors draw?

02Model

Everyone starts with the same picture: a robot is a machine that moves, so moving is where the energy goes, and the computer is a rounding error. On an autonomous mobile robot running its own navigation stack and object detector, the motors draw 16.6 percent of system power and the GPU alone draws 37.3 percent. The compute subsystem dissipates more than the sensors and motors together. Drive the same robot by hand with navigation and detection switched off and the motors jump to 64 percent. Nothing about the chassis changed. The workload changed. Autonomy is the case Physical AI means, and in that case the computer is the larger load.

03Guided practice

Run the breakdown cell for all four scenarios and read the motor share. Then open the live instrument, select AMR · autonomous, and confirm the same split. Switch the measurement plane to the motor rail and watch what happens to a hidden compute load.

04Feedback

PASS WHEN You measured the ratio and it runs the other way.

If your run fails, check:

Under autonomy the GPU/motor ratio is about 2.2 — compute is the larger term, not the smaller.
The same robot driven manually with autonomy off puts motors at 64%. The chassis is identical; only the workload moved.
This is why 'hidden computation is masked by actuation' is a claim about a workload, not about robots.

05Retrieve & space

Under autonomous control, what share of an AMR's system power do the motors draw?
Which changed between scenario 1 and scenario 4 — the hardware or the workload?
Why does the manual-driving scenario give a misleading picture of a Physical AI system?

06Mastery & project

State the compute-to-actuation ratio of a platform with the workload attached, and explain why quoting it without the workload is meaningless.

Find a robot power figure quoted in a datasheet or paper. Determine whether the workload is stated. Note what you cannot conclude without it.

GO FURTHER · THE LIVE PLAYGROUND

Open the instrument ↗ · Three bodies, two measurement planes. Watch the same hidden work appear and vanish.

← Energy as Evidence: What a Machine Can Prove About Itself Which plane can see it →

PYTHON · NUMPY

The body decides what the joules reveal

The four measured scenarios are given. Your job is to compute the compute-to-actuation ratio in each and say which one describes an autonomous robot.

You read

the arrays and values already in scope

You change

the code you write in each cell

Fixed

the dataset and the checks that grade you