Lesson 21 of 21 in this unit
Transfer a policy to a richer embodiment.
Compare planar velocity and 3D-engine torque contracts with explicit missing evidence.
By the end: Compare planar velocity and 3D-engine torque contracts with explicit missing evidence.
Prerequisites and path options
Why learn this? Identify the units, dynamics and feedback differences that prevent transferring a policy by merely copying its outputs.
Start with something concrete.
A command to move at half a metre per second is not a torque command. Changing the number’s label cannot make an actuator interface compatible.
Try the interactive modelDoes success in one simulator prove physical robot success?
Change one thing. Watch what follows.
Loading this browser experiment…
Understand what changed.
A shorter explanation, with the original worked examples available below. Both explain the same experiment and its limits.
The question
A command to move at half a metre per second is not a torque command. Changing the number’s label cannot make an actuator interface compatible.
Try it and inspect what changed
Inspect the contract table, then run or review the downloadable CPU simulator bridge.
Create an embodiment comparison covering observations, action units, sampling interval, termination, contact, gravity and omitted effects. Use the studio’s checks to reject malformed actions. Record the package versions and seed with your report so another person can replay the result.
The calculation in words
The planar policy outputs velocities in m/s from exact rendered positions. The supplied MuJoCo arm interface observes joint angles in radians and velocities in rad/s and accepts torques in N m. A downstream controller must translate goals into admissible torques; the browser’s output cannot be copied directly.
Keep the conclusion within the evidence
The bridge reuses a documented standard simulator, pinned dependencies, a proportional baseline and a trained imitation policy. It adds independent sampled execution checks and requires reporting all scenarios. The learner exercise asks for actual runnable code, tests and a changed scenario; the supplied reference is not the learner’s implementation.
The 3D engine still uses a simplified two-joint arm and sampled checks, not a swept-volume proof. No physical robot was tested by the browser lesson. Independently reviewed implementation evidence and representative learner pilots must be recorded by actual people.
You can explore every lesson directly. Completing this page, submitting a project report and demonstrating understanding on a separate case are different records. Optional code is available in the robot-learning studio; it does not increase guided lesson completion requirements.
Open the full worked examples, calculations and limitations
This is the same reference used in Technical mode. Expand it whenever you need the exact numbers, assumptions or further explanation.
Start with the task
A command to move at half a metre per second is not a torque command. Changing the number’s label cannot make an actuator interface compatible.
Mechanism and units
The planar policy outputs velocities in m/s from exact rendered positions. The supplied MuJoCo arm interface observes joint angles in radians and velocities in rad/s and accepts torques in N m. A downstream controller must translate goals into admissible torques; the browser’s output cannot be copied directly.
The bridge reuses a documented standard simulator, pinned dependencies, a proportional baseline and a trained imitation policy. It adds independent sampled execution checks and requires reporting all scenarios. The learner exercise asks for actual runnable code, tests and a changed scenario; the supplied reference is not the learner’s implementation.
Work through the evidence
Create an embodiment comparison covering observations, action units, sampling interval, termination, contact, gravity and omitted effects. Use the studio’s checks to reject malformed actions. Record the package versions and seed with your report so another person can replay the result.
Inspect the contract table, then run or review the downloadable CPU simulator bridge.
What this experiment can establish
The 3D engine still uses a simplified two-joint arm and sampled checks, not a swept-volume proof. No physical robot was tested by the browser lesson. Independently reviewed implementation evidence and representative learner pilots must be recorded by actual people.
Keep collection, fitting, frozen execution and interpretation as separate records. Save the configuration and complete result, including failures, in the robot-learning project. Report submission checks required evidence and text presence; reasoning and implementation quality need independent review.
Convert 0.6 m/s to mm/s, then consider a torque interface. Which statement is correct?
Take this idea with you.
The 3D engine still uses a simplified two-joint arm and sampled checks, not a swept-volume proof. No physical robot was tested by the browser lesson. Independently reviewed implementation evidence and representative learner pilots must be recorded by actual people.
Explore related topics and the topic map
How this topic connects
Explore this topic on the interactive mapUse these connections to review a foundation or choose where to go next. You can open any topic without completing the others.
What it builds on
- Geometry and control
Prevent coordinate-frame mistakes when combining robot positions, sensor readings and targets.
Compare planar velocity and 3D-engine torque contracts with explicit missing evidence.
What it enables
- Robot-learning project Apply in a project
Combine demonstration provenance, policy comparisons and complete-task evidence in the robot-learning project; keep embodiment limits and review status explicit.