Speed and separation monitoring is a collaborative operation mode in which a robot system continuously measures the distance to nearby people and adjusts its own speed so that it can always stop before contact. In robot safety, it is one of the four collaborative modes defined for industrial robots, alongside safety-rated monitored stop, hand guiding and power and force limiting.
The principle is simple: separation replaces fencing. Instead of a fixed guard, sensors watch the shared space and a safety controller enforces a protective separation distance that grows with robot speed and human approach speed. When a person comes closer than that distance, the robot slows down or performs a protective stop, and it resumes automatically once the person leaves.
How speed and separation monitoring works
Speed and separation monitoring works as a closed loop between safety-rated sensors, a safety controller and the robot drives. Sensors such as safety laser scanners, 3D safety cameras or safety radar detect people, the controller computes the current separation, and the robot receives a permitted speed or a stop command within each safety cycle.
The required distance is not a fixed number. It is calculated from the human approach speed, the robot speed, the reaction time of the whole system, the robot stopping distance and allowances for measurement uncertainty. The international guidance in ISO/TS 15066, now incorporated into the revised robot standards, expresses this as a sum of distance contributions.
The table below lists those contributions and what each represents.
| Contribution | What it represents | Main driver |
|---|---|---|
| Human motion | Distance a person covers before the robot stops | Approach speed, often 1.6 to 2.0 m/s |
| Robot reaction | Distance the robot moves before braking starts | Robot speed and system reaction time |
| Robot stopping | Distance traveled while braking | Payload, speed and stop category |
| Intrusion distance | Body parts reaching past the detection edge | Sensor detection capability |
| Position uncertainty | Measurement error of human and robot position | Sensor and encoder tolerances |
Because the robot term depends on speed, slowing the robot shrinks the required distance. This is what lets the robot keep working near people at reduced speed instead of stopping every time someone enters the cell.
Zones, speeds and safety functions
Most implementations divide the space into graduated zones rather than computing a single continuous limit. Static zones are simpler to validate, while dynamic separation recalculates the limit from the actual robot pose and speed, which allows tighter layouts but requires more capable sensors and controllers.
| Zone | Robot behavior | Typical use |
|---|---|---|
| Free zone | Full programmed speed | No person within the monitored area |
| Warning zone | Safety-rated reduced speed | Person approaching the cell |
| Protective zone | Protective stop with standstill monitoring | Person within the separation distance |
| Restart condition | Automatic resume after zone clears | Person has left the protective zone |
Every function in this chain must be safety-rated. The speed limit must be enforced by a safety-rated speed monitoring function, the stop by a monitored stop function, and the sensors must meet the required performance level. For collaborative robot cells, risk assessments commonly arrive at performance level d with category 3 architecture under ISO 13849-1, although the actual requirement follows from the specific risk assessment.
How it compares with other collaborative modes
Speed and separation monitoring is the only collaborative mode that aims to prevent contact entirely while the robot keeps moving. The other modes either stop the robot, let a person guide it, or accept contact within biomechanical limits.
| Mode | Contact allowed | Robot motion near people |
|---|---|---|
| Safety-rated monitored stop | No | Robot stands still while a person is inside |
| Hand guiding | Yes, through a guiding device | Robot moves only under operator control |
| Speed and separation monitoring | No | Robot moves at speed limited by distance |
| Power and force limiting | Yes, within thresholds | Robot moves with limited force and pressure |
The modes are often combined in one cell. A heavy industrial robot may use separation monitoring for normal work and switch to a monitored stop at the loading station, while a lightweight cobot may use power and force limiting at close range and separation monitoring to allow higher speeds when nobody is nearby.
Designing a cell with speed and separation monitoring
Designing a cell starts with the risk assessment and a measured stopping performance, not with sensor selection. The integrator measures stopping times and distances for the real robot, payload and speeds, then sizes the zones and places sensors so that no approach path bypasses detection.
- Measure stopping behavior at the worst-case payload, reach and speed rather than relying only on catalog values.
- Cover every access path, including approaches from below, from above and through gaps between machines.
- Account for shadowing, where the robot or parts block the sensor view and hide a person.
- Validate the full reaction chain, from detection through the safety controller to the drives.
- Plan restart logic so that automatic resume is permitted only when the risk assessment allows it.
Productivity depends on the size of the warning zone. Oversized zones make the robot slow down whenever someone walks past, while undersized zones risk violating the separation distance. Good layouts route foot traffic away from the robot and keep people outside the protective zone during normal work.
Limitations and common pitfalls
Speed and separation monitoring protects only against hazards that distance can control. It does not address sharp tools, ejected parts, hot surfaces or a workpiece swinging outside the robot envelope, and these hazards need separate measures.
| Problem | Cause | Remedy |
|---|---|---|
| Frequent nuisance stops | Zones drawn too large | Recalculate with measured stopping data |
| Undetected approach | Sensor shadow or blind spot | Add sensors or change mounting height |
| Separation violated in tests | Stopping distance underestimated | Reduce speed or enlarge zones |
| Unsafe automatic restart | Restart not tied to zone status | Require a clear zone or manual reset |
Changes after commissioning are a frequent source of risk. A heavier gripper, a new program with faster moves or a relocated scanner can all invalidate the original calculation, so any such change should trigger a renewed check of the separation distance.
A plain-language summary
Speed and separation monitoring lets a robot and a person share a space without a fence. Sensors watch how close the person is. When the person is far away, the robot works at full speed. As the person comes closer, the robot slows down, and if the person gets too close, the robot stops before it could touch anyone.
It works like a careful driver in a parking lot who slows down when a pedestrian approaches and brakes if they step in front of the car. Once the person walks away, the robot simply continues its work.
An example from practice
A packaging plant installs a palletizing robot at the end of a production line. Workers need to remove full pallets and bring empty ones several times per hour, and a fenced cell with interlocked doors would force a full stop each time.
The integrator mounts two safety laser scanners at ankle height and defines a warning zone and a protective zone around each pallet station, sized from measured stopping distances with the heaviest layer of boxes. When a worker enters the warning zone, the robot drops to a safety-rated reduced speed; in the protective zone it stops on the side the worker is using while continuing on the other pallet. The plant removes the fence on the access side and keeps production running during pallet changes.