Operator Curriculum

A humanoid under software control is never truly at rest. Residual torque in harmonic drives, capacitor charge in motor controllers, and stored energy in lift counter-balances can move a robot after power is cut.

The LOTO procedure exists to make the zero-energy state visible, verified, and non-reversible for the duration of the work — not just to cut power.

1. Notify: tell the fleet operator and any nearby staff that the robot is going into service. Log an out-of-service event in TechMedix.

2. Isolate: disconnect the primary power source (battery cut-off or mains breaker). Do not rely on soft shutdown alone.

3. Lock: apply a personal lock to the isolation point. Each technician working on the robot applies their own lock.

4. Tag: attach a tag with your name, date, and the fault being serviced.

5. Verify: attempt to operate the robot via its normal controls. A failed start confirms isolation.

6. Release: only after all work is complete, all locks are removed (each technician removes their own), and the robot is cleared for power-up.

Mechanical work (joints, covers, frame): safety glasses, cut-resistant gloves, steel-toe boots.

Electrical work (battery, controllers, wiring): arc-rated gloves, face shield, insulated tools, Class 0 gloves for >50 V.

Battery handling: chemical-resistant gloves and eyewear when inspecting swollen or vented cells.

Structural first: frame integrity, joint play, skin damage, fastener torque. A loose joint cap can mask a failing bearing.

Electrical second: connector seating, cable chafe, ground strap integrity, dust and moisture ingress around battery compartments.

Software last: status LEDs, ping response, TechMedix alert history. Software symptoms often point back to structural or electrical causes.

Harmonic drive: grease leakage around the output hub indicates seal failure. Flaking on the wave generator = end of life, retire immediately.

Cable runs: shiny wear spots on the jacket indicate rubbing against a moving surface. Replace before the conductor is exposed.

Skins and covers: compression sets (permanent dents in foam or rubber) reduce impact protection; cracked panels transmit more vibration into sensors.

A — new or near-new condition, no remedial work needed

B — normal wear, schedule routine service

C — advanced wear, service within 30 days

D — defect found, remove from service

Cell voltage spread: a spread greater than 50 mV across the pack under load indicates imbalance or a weak cell.

Temperature delta: a single cell more than 5 °C hotter than its neighbors is a leading indicator of internal resistance rise.

State of Charge (SOC) vs. State of Health (SOH): SOC is what's in the pack right now; SOH is how much of the original capacity remains. Retire at SOH < 80%.

Charge at ambient 10–30 °C. Charging a cold-soaked pack (<0 °C) causes lithium plating and permanent damage.

Store at 40–60% SOC for any period longer than 72 hours. Full-charge storage accelerates calendar aging.

Packs removed from service go into labeled, fire-rated storage until disposition (return, warranty claim, or disposal).

1. Get people away first. A venting pack releases toxic and flammable gases.

2. Cut power upstream — do not try to disconnect the pack directly.

3. Contain, do not extinguish. Li-ion fires re-ignite; cool with large volumes of water or isolate in a fire-safe container.

4. File an immediate safety event in TechMedix and notify the fleet operator and the local fire authority per site protocol.

Critical: safety, fire, or loss-of-service risk. Immediate technician dispatch. Escalate to operator within 15 minutes if not acknowledged.

Warning: degraded performance, scheduled intervention. Resolve within the current service window.

Info: state change, no action required unless patterned. Useful for trend analysis.

Acknowledging an alert tells the system a human has seen it; it does not mark it resolved.

Resolve only after the fault is fixed, validated, and the closing artifacts (photos, readings, protocol ID) are attached.

Never acknowledge an alert you did not investigate — the audit log records who clicked and when.

L1 to L2: any mechanical repair deeper than cover-off inspection, any electrical work above the BMS interface, any firmware update that fails validation.

L2 to L3: any fault observed on more than one robot of the same model within 7 days, any diagnostic that requires multi-platform context.

L3 to L4+: any systemic issue affecting fleet uptime SLA, any root cause that implicates platform definition or firmware baseline.

Configuration backup exported to TechMedix and locally archived.

Firmware checksum verified against the published release notes.

Robot on stable mains power or a pack at >80% SOC — never update on a degraded pack.

Clear work envelope around the robot in case of reboot to an unknown pose.

Run the full range-of-motion sequence and compare joint currents to the pre-update baseline.

Confirm all sensors are reporting within expected noise floor.

Verify TechMedix heartbeat and that the new firmware version is recorded in the robot's asset record.

If any check fails, roll back to the previous firmware using the saved backup and escalate.