Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for automation engineers

Motion Tuning Loop Agent

Reach the target response with the fewest runs and without leaving safe limits

DecisionsAutomationIT and development
See how it worksGet access
Motion Tuning Loop Agent: what goes in, what the agent does and what you get

What it does

Tuning a drive by hand means changing a gain, running a move, looking at the trace and guessing again. This agent runs that loop under safe limits. It runs a step or trajectory test, reads the response data, and computes overshoot, rise time, settling time and steady-state error. It compares them with the targets, proposes new gains using the pattern in the response, and runs again. Every run stays inside limits for current, speed, position and vibration, and the agent aborts if one is crossed. It stops when the targets are met or the run budget is used. The engineer approves any value written to the production controller. Edge case: higher gain cuts settling time but excites a resonance, so the agent backs off and adds a notch filter option.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
Yes, continueYes, continueApprovedNoNo 1 STARTS WHEN Tuning session requested 2 USES A TOOL Confirm safe limits and clear the work area 3 USES A TOOL Run the step or trajectory test and record responsedata 4 DOES Compute overshoot, rise time, settling time anderror 5 CHECKS THE RESULT Were all safety limits respected during the run? If not: abort, lower the gains to the last safe set, andrestart. Back to step 2. 6 CHECKS THE RESULT Are all targets met? If not: propose new gains from the response shape andrun again, within the run budget. Back to step 3. 7 USES A TOOL Run a repeat test with a different load or move toconfirm 8 YOU APPROVE Engineer approves writing gains to the productioncontroller 9 USES A TOOL Write gains and record the old set for rollback 10 RESULT Tuning report with gains and response plots
Read the steps as a list
  1. Tuning session requested
  2. Confirm safe limits and clear the work area
  3. Run the step or trajectory test and record response data
  4. Compute overshoot, rise time, settling time and error
  5. Were all safety limits respected during the run?If not: abort, lower the gains to the last safe set, and restart. Back to step 2.
  6. Are all targets met?If not: propose new gains from the response shape and run again, within the run budget. Back to step 3.
  7. Run a repeat test with a different load or move to confirm
  8. Engineer approves writing gains to the production controllerThe agent waits here for your OK.
  9. Write gains and record the old set for rollback
  10. Tuning report with gains and response plots

How it decides

It adjusts one or two gains per run by an amount based on the measured error and stops increasing a gain as soon as vibration or overshoot passes the limit.

  • Abort any run when vibration or current crosses its limit
  • Reduce the proportional gain by 20% when overshoot exceeds the target
  • Stop raising a gain when a resonance peak appears
  • Limit the session to 25 runs

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Targets for overshoot and settling
  • Safe limits per axis
  • Run budget (default 25)
  • Test move type
  • Rollback gain set

What keeps you in control

It always asks you first

  • Engineer approves writing gains to production
  • Safety lead approves limit changes

Hard limits

  • Never exceed the safe limits to speed up tuning
  • Never write gains to production without approval

It stops when

  • Done: targets are met on two different tests
  • Stop: the run budget is used or a mechanical issue is found

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensA servo axis needed under 5% overshoot and 80 ms settling. Run 1 gave 14% and 120 ms. The agent cut gain and raised damping. By run 7 it had 4% and 76 ms, but the confirmation test with a heavier load rang at 220 Hz, so the check failed. It added a notch filter at 220 Hz and the retest gave 3% and 82 ms. The engineer approved it, accepting 82 ms.

More agents for automation engineers

Quality control · IT and development

Runbook Freshness Test Agent

Runbooks that are tested, current and dated

10 steps1 check7 jobs
Quality control · IT and development

Hardware-in-the-Loop Test Run Agent

Run the full rig suite unattended and classify each failure with evidence

10 steps2 checks8 jobs
Quality control · IT and development

As-Built Wiring Verification Agent

Make the drawings match the installed wiring before startup

11 steps1 check3 jobs
Monitoring · IT and development

Controller Program Version Compare Agent

Keep the repository equal to what really runs, with every difference explained

11 steps2 checks4 jobs