RobotGuide

Engineering practice · topic F7 · 20 hours

Functional safety design

A robot is «a machine that can hurt people». Safety design is not bolting on an e-stop button; it is a state machine plus failure analysis.

Engineering stage · Hardcore
System software · Proficient Embedded & controls · Proficient Mechanical · Hands-on

Tools
Tools

What to learn

  • Safety state definitions: power-off, brake, retract, speed limit — when each one triggers
  • E-stop chain and watchdogs: if the software dies, can the hardware still stop it
  • Failure mode analysis (FMEA): the single-point failure list
  • Four classes of protection: limits, timeouts, overspeed, overcurrent

The lesson · plain words

Functional safety answers one question: «what if the machine hurts someone». It is not as simple as adding an e-stop button — it is a complete process from risk analysis to verification. At entry level you only need one diagram: the risk matrix.

The risk matrix: where all safety design starts

Place every hazard scenario on a severity × exposure-frequency matrix: severe and frequent = a safety function must be designed. Drag the slider below to feel it: the same robot in a busy workshop (frequent) and in a lab (occasional) demands completely different safety levels.

The risk matrix result maps to ISO 13849 performance levels PL (A–E): collaborative robots typically require PL=d. Higher levels demand more redundancy in the safety loop and stricter verification — with costs differing by an order of magnitude.

Interactive figure · drag it and see

What this figure shows: a severity × exposure grid graded QM/low/mid/high across minor→catastrophic severities and very rare→frequent exposure, with a movable scenario cell and a live risk-level readout that names the required response. Takeaway: functional safety starts by putting every «the robot could hurt someone» scenario on this chart; frequent exposure pushes a moderate risk to high and a safety function becomes mandatory.

The risk-reduction trio, in order

① Design it out (top priority): reduce speed and force so collision becomes harmless — a collaborative arm's «power and force limiting» is exactly this idea; ② guard it: fences, light curtains, and safety zones keep people out of the danger; ③ procedures and training: the least reliable layer, good only as a backstop, never the primary defense.

Safety loop design principles: the safety function is independent of the main control (independent sensors + independent logic + independent disconnect path), verifiable (regular self-tests), and fail-safe (a broken wire = stop, not keep running).

A robotics-specific hazard checklist

High-frequency scenarios: hand pinching (gripper force limiting), collision injury (speed and force monitoring, safety-zone scaling), falling (power-off brake), runaway motion (software + hardware dual limits), battery thermal runaway (temperature monitoring + circuit breaker).

Every entry goes through the risk matrix and gets documented: scenario, severity, frequency, measure, verification method. In a customer audit or an accident investigation, this document is your only amulet.

How deep to go

You can produce a complete hazard identification list for a robot and place it on the matrix; explain the relation between PL levels and redundancy; and design one fail-safe loop. This is hard-threshold knowledge for entering industrial robotics — and the easiest place to stand out in an interview.

Three questions — do you really get it?

1. What is the priority order of risk-reduction measures?

  • ADesign it out > guard it > procedures and training
  • BTraining > guarding > design
  • CAll in parallel
Show answer

A. Eliminating at the source is the most reliable; human behavior is the least reliable. Training is only a backstop.

2. What is the design requirement of «fail-safe»?

  • AKeep running through the failure
  • BAny failure leads to a stop / safe state, e.g. a broken wire = stop
  • CNever fail
Show answer

B. The design axiom of safety loops: assume failure will happen, and the post-failure state must be safe.

3. For the same robot, what does higher exposure frequency mean?

  • ALower risk
  • BIt does not matter
  • CA higher risk level on the matrix, stricter safety requirements
Show answer

C. Frequency and severity jointly determine risk. Frequent exposure pushes «moderate» up to «high».

Keep going deeper

Links without a marker stay on this site; the rest point to the Chinese edition until its translation batch lands.

How it fails in the field

  • Software limits only; the first encoder failure drives it straight into a hard crash
  • E-stop cuts motor power without braking; a heavily loaded joint keeps falling under gravity

You have learned enough when

You can draw the safety state machine of a robot, list five single-point failures, and give a countermeasure for each.

The Chinese edition carries the full topic page for Functional safety design, including the complete outline, depth requirements per career track, and further derivations: robotguide.dev/k/F7/.