A collaborative robot installed beside an operator is not automatically a fence-free workstation. When production teams ask, do cobots need guarding, the accurate answer is: sometimes. The cobot’s safety functions may reduce the need for traditional perimeter fencing, but the finished application must still be assessed for the risks it creates.
That distinction matters commercially as well as legally. Removing unnecessary guarding can save floor space, simplify loading and make automation easier for operators to use. Removing safeguarding where it is required can create an unsafe cell, repeated protective stops and an installation that cannot be put into production with confidence.
Do cobots need guarding in every application?
No. A cobot can often work without a conventional cage when the risk assessment shows that people can share the workspace without unacceptable injury risk. This may be practical for light pick-and-place, inspection, assembly or machine tending tasks where the robot moves at limited speed, carries a benign workpiece and has no hazardous tooling.
But “collaborative” describes the robot’s available operating modes and safety-related capabilities. It does not certify every combination of robot, gripper, product, fixture and process as safe for direct human contact. A 20 kg payload robot moving a metal casting, for example, presents a very different risk from a compact arm transferring empty plastic trays.
The decision is made at workstation level. The responsible party must assess the complete system, including the robot, end effector, fixtures, infeed and outfeed equipment, controls, process hazards and foreseeable operator behaviour. Applicable machinery legislation and relevant standards, including ISO 10218 and ISO/TS 15066, provide the framework. The final safety concept should be validated for the actual task rather than assumed from a product brochure.
What determines whether physical guarding is required?
Risk assessment starts with the question: what could hurt someone, and under what conditions? Collision is only one possibility. In many installations, the greater risk comes from the process around the robot.
A force-controlled cobot may safely stop when it contacts a person, yet the same installation can still have trapping points between the arm and a fixture. A screwdriving application can introduce rotating tooling and sharp fasteners. Welding brings arc radiation, fumes, hot parts and spatter. Palletising can involve unstable loads, while machine tending may expose the operator to sharp components or a machine cycle that has its own safeguarding requirements.
The most influential factors are usually the robot’s speed and momentum, payload, reach, tool geometry, workpiece characteristics and the accessible pinch or crush zones. The frequency of human access matters too. If an operator only enters a cell during planned changeovers, a gated enclosure may be the clearest and most productive solution. If they need to load a part every 30 seconds, a carefully designed collaborative workspace may be more appropriate.
The intended operating mode must also be considered. A system may be safe for hand guiding during teaching at reduced speed, but not when it returns to automatic production. Equally, a low-speed collaborative path might be acceptable near the loading position while faster motion can take place in a separated area.
Four ways a collaborative workspace can be safeguarded
Collaborative operation is not one fixed setup. The usual safety strategies are power and force limiting, speed and separation monitoring, safety-rated monitored stop, and hand guiding. More than one strategy is often used in the same workstation.
With power and force limiting, the robot is designed to limit contact forces and stop when set thresholds are exceeded. This can support direct human-robot interaction, but it does not make every contact point acceptable. The assessment must examine body regions, contact duration, potential impact speed and whether a person could be trapped against a fixed object.
Speed and separation monitoring uses safety-rated sensing to manage distance between people and the robot. As a person approaches, the robot slows, then stops before the separation distance is breached. This can preserve cycle time better than running the entire process at a low collaborative speed. It also demands disciplined layout design: scanner fields, approach routes, blind spots and restart behaviour must all be tested in the real production environment.
A safety-rated monitored stop keeps the robot stationary while an operator enters the shared space. Once they leave, the system can resume under defined restart conditions. It is often a sensible approach for inspection, loading or part verification where true simultaneous working is unnecessary.
Hand guiding is primarily useful for teaching paths or positioning the arm manually. It should not be treated as a substitute for safeguarding an automatic process.
Physical guards, interlocked doors, light curtains and safety scanners remain valid tools. The strongest solution is often hybrid: fixed guarding around the rear and sides, a protected operator access point at the front, and reduced robot speed close to the shared area.
Why tooling changes the answer
The robot arm is only one part of the hazard calculation. End-of-arm tooling frequently decides whether an open collaborative layout is realistic.
A rounded vacuum gripper carrying a lightweight carton may be suitable for close operator access after validation. A pneumatic gripper with sharp jaws may not be, even if the robot itself has sensitive collision detection. Tool movement can create shearing points. Loss of air or power can drop a part. A hot, oily, brittle or sharp workpiece adds risks that no force sensor can remove.
This is why application-specific design is more valuable than a generic claim that a cobot is safe beside people. The tooling should be selected and designed for the handling task, with workpiece retention, safe states and accessible maintenance considered from the start. Fixtures need the same attention. A harmless robot motion can become hazardous when it pushes a hand against a hard locating pin or machine frame.
Guarding can improve productivity, not just safety
A common mistake is to see guarding as a choice between maximum safety and minimum cycle time. In practice, an open cell that slows the robot every time someone walks past can deliver less output than a compact guarded cell with a well-designed loading station.
For high-speed repetitive work, separating people from fast robot motion may produce the most predictable throughput. Operators can focus on preparation, quality checks and replenishment instead of repeatedly waiting for protective stops. The system becomes easier to troubleshoot because its permitted operating states are clear.
Conversely, full fencing can be counterproductive for low-volume, high-mix work that requires frequent intervention. Here, an accessible cobot setup with appropriate speed limits, restricted motion zones and simple operator controls can reduce handling effort without consuming excessive floor space.
The objective is not to eliminate fences at all costs. It is to match the safety architecture to the process, labour pattern and expected return on investment.
A practical assessment before specifying the cell
Before selecting safeguards, map the actual cycle. Observe how parts arrive, where operators stand, how they clear faults, which tools they use and what happens during changeover, cleaning and maintenance. The normal cycle is only part of the picture. Many incidents occur while recovering from a mispick, reaching for a dropped component or adjusting a fixture.
Then test the proposed layout against credible scenarios. Could a person be caught between the robot and a table? Can they enter from an unexpected side? What happens if a part slips? Does the robot restart only after a deliberate action? Are safety devices suitable for dust, reflections, welding light or congested aisles? These questions turn a theoretical safety concept into a workable one.
For FAIRINO US projects, this stage is where feasibility assessment and workstation design add practical value. A live test with the real part, gripper and intended motion can reveal whether collaborative operation is genuinely suitable, or whether partial guarding will provide a safer and more productive result.
When a conventional industrial cell is the better choice
There are applications where a cobot should be guarded, and others where a different automation approach is the better engineering decision. High speeds, heavy payloads, long reaches, sharp tooling, uncontrolled loads and hazardous processes usually push the design towards physical separation.
That is not a failure of collaborative robotics. It is a sensible recognition of application boundaries. A guarded robot cell can still be compact, straightforward to operate and economically justified. The right question is not whether a fence looks less advanced. It is whether the completed workstation allows people to work safely while delivering the cycle time, quality and uptime the business needs.
Start with the task rather than the robot label. A well-tested risk assessment will show where open collaboration adds value, where selective safeguards are enough, and where a guarded layout gives your team the confidence to run production at pace.