📋 Pre-Commissioning Risk Assessment
Before any robot goes into operation, it is required by standard carry out a systematic process of hazard identification and risk assessment. The standards ISO 12100 (general principles) and ISO 10218 require this step: without it, the cell cannot be commissioned.
🎯 What It Is and Why It Matters
The risk assessment is the baseline of everything that follows: it defines safety zones, maximum speeds, and permitted contact forces. Without it documented, the company operates out of compliance, with legal and accident risks.
- •ISO 12100 — general principles for machine safety.
- •HIRA — Hazard Identification and Risk Assessment.
- •Severity × probability matrix — classifies each hazard.
- •Mandatory documentation — with no record, there was no evaluation.
✓ What to DO
- ✓Execute the HIRA before during commissioning, with the real cell.
- ✓Document each hazard in the severity × probability matrix.
- ✓Derive zones, speeds, and forces from the assessment.
- ✓Review the assessment whenever the layout or tools change.
✗ What NOT to do
- ✗Copying an assessment from another “similar” cell.
- ✗Commission first and “document later.”
- ✗Relying only on the manufacturer's spec sheet without evaluating real-world use.
- ✗Treating the document as filed-away bureaucracy that’s never consulted.
💡 Practical Tip
Treat the severity × probability matrix as a living document: every near miss and every newly integrated tool should trigger a review. An assessment from 2 years ago probably doesn’t reflect today’s cell.
💪 Contact Force Validation
Measurement physical of the forces applied by a collaborative robot, validating them against the biomechanical limits of the ISO/TS 15066. Real measurement, not theory: manufacturers publish rated forces, but with tools and parts attached, the actual force can exceed the limits.
📊 Data you need to measure
- ISO/TS 15066 — sets force and pressure limits for each body region.
- Transient vs quasi-static — brief contact (impact) versus prolonged contact (crushing) have different limits.
- CBSF — biomechanical force measurement devices used in situ.
- Recalibration — frequency is defined; validation isn’t a one-time event.
⚠️ Warning: rated force is not actual force
The force rating published by the manufacturer applies to the robot nu. When you attach a gripper, tool, and workpiece, mass and geometry change — the in-situ contact force can exceed the limits of ISO/TS 15066. On-site validation is mandatory and must be redone every time the end-effector is changed.
💡 Practical Tip
Measure the force in the body regions most exposed in the actual application (hand, forearm, torso) and at the points along the trajectory where speed is highest. Always distinguish transient contact from quasi-static contact—the limits are not the same.
🔄 SSM/PFL and Collaborative Modes
A ISO 10218-2 defines 4 collaborative operation modes. Each application requires a specific mode — using the wrong mode is as dangerous as having no protection at all. It is PFL no is synonymous with "safe in every situation."
Safety-rated Monitored Stop
Safety-rated monitored stop
The robot stops when a human enters the zone and resumes only when they leave. Coexistence without simultaneous movement.
Hand Guiding
Manual guide
The operator physically moves the robot using a guiding device. Movement occurs only under direct hand control.
Speed and Separation Monitoring (SSM)
Speed and separation monitoring
Sensors maintain a minimum distance; the robot slows down as the human approaches and stops if the separation falls below the limit.
Power and Force Limiting (PFL)
Power and Force Limitations
The robot limits force and power to biomechanically safe levels, allowing contact. It depends directly on validation under ISO/TS 15066 (topic 2).
⚠️ Attention to mode
Each application requires a specific mode. Using the wrong mode is as dangerous as having no protection at all. PFL does not mean “safe in every situation” — sharp edges and piercing tools can cause injury even within force limits. Modes can be combined and must transition in a controlled manner.
🔒 LOTO Adapted for Robots
The classic Lock-Out Tag-Out needs to be adapted for robotic cells. Robots have gravitational potential energy on the axes: cutting the power doesn't mean the arm won't fall. LOTO for robots requires additional procedures that classic LOTO doesn't cover.
✓ Robotic LOTO (correct)
- ✓Consider residual energy and the gravitational potential of the axes.
- ✓Engage the mechanical brakes before entering the zone.
- ✓Verify zero energy (explicit confirmation procedure).
- ✓Using a specific tag for the teaching/programming mode.
✗ LOTO Classic applied without adaptation
- ✗Assume that cutting the power locks the arm.
- ✗Ignore the vertical axis falling due to gravity.
- ✗Skip the zero-energy verification.
- ✗Enter the cell in teaching mode without a dedicated tag.
De-energize and isolate
Cut off the energy source
Isolate and lock out the cell’s electrical, pneumatic, and hydraulic power sources with a lock and tag.
Lock axes with a mechanical brake
Contain gravitational energy
Engage the mechanical brakes to prevent the vertical axes from falling under load, which cutting power alone cannot prevent.
Verify zero energy
Confirm before entry
Explicitly confirm that no residual energy is stored before any part of the body enters the zone.
Teaching mode tag
Specific indication
When the cell must be powered on for programming, use a dedicated tag that signals active teaching mode.
💡 Practical Tip
The most dangerous axis is the vertical one under load: when power is cut, gravity tends to pull it down. Always engage the mechanical brake and confirm zero energy before any body part from entering the danger zone. The teaching-mode tag should look visually different from the maintenance tag.
🚑 Incident Response
A robot touches a worker: what now? Is there a standardized sequence of actions for unplanned contact. The first 60 seconds determine everything: injury severity, preservation of evidence for investigation, and legal responsibility.
E-stop
Immediate emergency stop
Press the emergency stop to halt any robot movement before approaching.
Provide aid
Worker support
Provide first aid and call for medical assistance. The person’s life comes before any record.
Isolate
Preservation of evidence
Isolate the area and preserve the cell state and controller logs for the root cause investigation.
Record and notify
Mandatory notification
Record what happened and notify as required (CIPA/SESMT), initiating the root cause investigation.
🚨 The first 60 seconds
Do not erase or restart the controller before preserving the logs: they are the central evidence in the investigation. The chain E-stop → provide assistance → isolate → document must be trained and posted in the cell — no one improvises the response sequence in the middle of an accident.
📝 Near-Miss Culture
A reporting system for near misses without punishment. For every serious accident, there are about 300 near-misses not reported (Heinrich pyramid). Capturing near-misses prevents accidents before that happen.
📊 Heinrich's pyramid
- ~300 near-misses for each serious accident — these are leading indicators.
- Anonymous reporting — with no punishment, people report what really happened.
- Near-miss classification — enables trend analysis.
- Simulation of erratic human behavior (Springer 2024) — tests the cell's robustness.
✓ Healthy culture
- ✓Anonymous reporting with no punishment.
- ✓Classify and analyze near-miss trends.
- ✓Test the cell with simulated erratic human behavior.
- ✓Address recurring near-misses before an accident occurs.
✗ Toxic culture
- ✗Punish those who report near misses.
- ✗Only recording accidents that have already caused an injury.
- ✗Ignore near misses because they’re “without consequences.”
- ✗Assume that humans always act predictably.
💡 Practical Tip
A near miss is free: it’s an accident that cost nothing and still gave you a warning. Make reporting as easy as possible (anonymous, in 30 seconds) and publicly show the actions taken — that’s what builds trust and keeps reports coming in.
🛡️ Module Summary
Next Module:
2.3 - 🥽 Simulators, VR, and Digital Twins for Training