May 13, 2026 · Standards Compliance
The cobot pitch in 2026 sounds simple. Collaborative robot. No fence required. Drop it on a workbench next to the operator. The marketing has done its job, and West Michigan auto suppliers, plastics processors, packaging plants, and food and beverage manufacturers have been adopting cobots at a steady pace through 2025 and into 2026.
The compliance reality is more complicated. A cobot is still an industrial robot under ANSI/RIA R15.06-2025 and ISO 10218-1:2025 and ISO 10218-2:2025. ISO/TS 15066 (Robots and robotic devices - Collaborative robots) is the technical specification that fills in what collaborative operation actually means under those main standards. A fenceless install without a documented risk assessment that meets 15066 is not a compliant install. We see this gap on most cobot audits we run in the Grand Rapids, Holland, Kalamazoo, and Battle Creek corridor.
This guide walks through what ISO/TS 15066 actually requires, the four collaborative methods, the biomechanical limits that define power and force limiting installs, and the documentation that holds up to MIOSHA and OSHA review.
A West Michigan cobot deployment falls under at least four standards simultaneously:
The four documents do not contradict. They stack. A cobot install must meet all of them. The risk assessment cited in OSHA enforcement of the General Duty Clause is the same risk assessment cited in ISO 10218-2 Section 6 and the same risk assessment that validates ISO/TS 15066 biomechanical limits.
ISO/TS 15066 defines exactly four collaborative operation methods. A cobot install must implement one or more of these to claim collaborative operation. Any operation outside the four methods reverts to standard industrial robot rules with the same fencing and access control requirements as a six-axis FANUC, ABB, KUKA, or Yaskawa weld cell.
The robot is stopped when the operator is in the collaborative workspace. Motion resumes only when the operator clears the workspace. Implemented with safety-rated soft stop functions in the controller, typically using PLd Cat 3 or higher safety circuits. Common on FANUC CRX and ABB GoFa applications where the operator and robot share the cell but do not work simultaneously.
The operator guides the robot through motion by applying force to a dedicated handle or enabling device. The robot moves only while the operator applies force. Less common in production; more common in teaching mode for path programming or in research applications. Requires a safety-rated handle (Universal Robots Polyscope offers this; KUKA iiwa has integrated hand-guiding capability).
External safety sensors (laser scanners, light curtains, 3D vision) monitor distance between the robot and the operator. As the operator approaches, robot speed reduces. At a defined separation distance, the robot stops. When the operator moves away, the robot resumes. Common on larger cobots and on cobots running near full-speed industrial throughput when the operator is not in the cell. SICK microScan3, Pilz SafetyEYE, and Keyence SZ-V series are the most-installed SSM sensors in West Michigan plants.
The robot itself is inherently safe through limited mass, speed, and contact force. The robot does not need to detect the operator because the worst-case impact is below biomechanical injury thresholds. Universal Robots (UR3, UR5, UR10, UR16e, UR20, UR30) and FANUC CRX series are the most common PFL cobots in West Michigan. Doosan, Techman, and Kassow also appear in the regional install base.
PFL is the method most West Michigan cobots deploy under. It is also the method that most frequently fails audit because of end-of-arm tooling changes that invalidated the original PFL validation.
Annex A of ISO/TS 15066 lists quasi-static and transient force and pressure limits for 29 specific human body regions. A PFL cobot installation must verify by measurement (not assumption) that the contact force and pressure at every reachable point on a worker's body is below these limits.
The values are research-derived from biomechanical pain-threshold and injury studies. They are not opinion. They are part of the standard. Selected values for free-arm contact (cobot freely moving in space, operator close to robot):
| Body region | Quasi-static force limit (N) | Transient force limit (N) | Pressure limit (N/cm²) |
|---|---|---|---|
| Forehead | 130 | 175 | 130 |
| Face | 65 | 90 | 110 |
| Neck | 150 | 190 | 140 |
| Shoulder | 210 | 320 | 160 |
| Chest (sternum) | 140 | 210 | 120 |
| Abdomen | 110 | 160 | 140 |
| Upper arm | 150 | 220 | 150 |
| Forearm | 160 | 220 | 180 |
| Back of hand | 140 | 190 | 200 |
| Palm of hand | 140 | 200 | 260 |
| Thigh | 220 | 320 | 160 |
Verification uses a calibrated biomechanical test device. The Pilz PRMS, Pilz Robot Measurement System (and equivalent units from PicoForce and FORCE-X) presses the cobot end-effector against a calibrated load cell at production speed and reports peak force and pressure. We perform this measurement on every PFL cobot we audit. The Universal Robots datasheet says the bare arm meets the limits. The bare arm sometimes does. The arm with a 4 kg pneumatic gripper holding a 3 kg machined aluminum part often does not.
Per ISO 10218-2 Section 6 and ISO/TS 15066 Annex C, the risk assessment for a collaborative cobot must include:
The file is what an inspector wants to see. Without the file, the install is presumed non-compliant under MIOSHA enforcement practice. With a clean file that traces every hazard through the selected method and the verification measurement, the install holds up.
The original risk assessment was performed on the bare cobot. The plant later added a gripper, suction tool, vision-guided pickup head, or dispenser. The risk assessment was not updated. About 70 percent of fenceless cobot audits we run have this gap.
The plant relies on the robot manufacturer's PFL datasheet. The datasheet is for the bare arm under specific conditions. The as-installed system with its tooling and payload was never measured. About 60 percent of audits have this gap.
ISO/TS 15066 governs collaborative operation but does not eliminate the need for OSHA 1910.147 LOTO during service and maintenance. We find cobot installations with no documented LOTO procedure, or LOTO procedures that omit residual energy (servo drive capacitor bleed-down time on UR robots is 4 seconds; on KUKA iiwa it is documented in the controller manual). About 55 percent of audits have a LOTO gap. See our Robot Cell LOTO standards alignment guide for the full LOTO walkthrough.
The cobot arm itself is rounded and PFL-compliant. The gripper jaws, the suction cup mount bracket, the screw heads on the end-effector flange are not. ISO/TS 15066 Annex A specifies pressure limits, which means a sharp edge can fail even when total force is below the limit. About 45 percent of audits flag this for re-engineering of tooling or for guards on the tool.
Workers near a collaborative cobot must be trained on the collaborative operation method, the emergency stop locations, the residual risk, and the response to abnormal robot behavior. Documentation is required under MIOSHA and OSHA general duty enforcement. We find this gap on about 40 percent of audits, especially in plants that rotate operators across multiple cells.
What we put in place on a typical greenfield cobot deployment for a West Michigan auto supplier or plastics processor:
The full stack typically runs 30 to 80 pages depending on complexity. Cells with multiple cobots, multiple methods, or shared collaborative workspaces run longer. For greenfield cobot deployments we provide the full stack as part of the robotics gap analysis service.
Re-verification is required (per ISO 10218-2 Section 7 and ISO/TS 15066 Section 5) on any of:
The re-verification process is shorter than the initial assessment (typically a half day for an experienced safety engineer plus the measurement step). The cost of not re-verifying is OSHA citation under the General Duty Clause and any worker injury that traces to an unaccounted-for hazard.
We will walk your cell, review the documentation, perform a PFL measurement check, and write a gap report. No high-pressure sales pitch, no overbilling for findings. Direct line: 616-217-3325.
Request Gap AssessmentNo. Collaborative robots are still industrial robots under ANSI/RIA R15.06-2025 and ISO 10218-1:2025 / ISO 10218-2:2025. ISO/TS 15066 is a technical specification that supplements the main standards with the specific safety requirements for collaborative operation. A cobot installation must still meet the general industrial robot system safety requirements, plus the additional collaborative requirements in 15066.
Safety-rated monitored stop, hand guiding, speed and separation monitoring, and power and force limiting. Only one or more of these four methods constitutes true collaborative operation. The first three rely on external safety hardware (light curtains, laser scanners, force-sensitive enabling devices). Power and force limiting relies on the robot itself being inherently safe through limited mass, speed, and contact force. Most cobot deployments in West Michigan plants use power and force limiting or speed and separation monitoring.
ISO/TS 15066 Annex A defines maximum permissible quasi-static and transient force and pressure values for 29 specific human body regions. For a free-arm contact (typical cobot interaction), the quasi-static force limit on the back of the hand is 140 N, on the face is 65 N, on the chest is 140 N. Transient (impact) limits are roughly double the quasi-static. These values must be verified by measurement on the actual installed cobot, not assumed from the robot manufacturer's datasheet.
Sometimes, after a proper risk assessment confirms it. A fenceless cobot installation requires documented risk assessment per ISO 12100, validated against ISO/TS 15066 biomechanical limits, with the collaborative operation method explicitly defined. End-of-arm tooling (gripper, weld torch, suction cup) changes the math significantly. A bare cobot may pass biomechanical limits. The same cobot with a sharp-edged gripper and a 5 kg part in the gripper often does not.
The integrator and the end-user share responsibility under ISO 10218-2 and ISO/TS 15066. The integrator typically performs the initial risk assessment as part of system handoff. The end-user (the plant) must verify the risk assessment matches the actual installed conditions, perform a periodic re-assessment when anything changes (tooling, application, part geometry, environment, throughput), and maintain documentation evidence under OSHA 1910.147 and applicable Michigan MIOSHA records retention rules.
End-of-arm tooling not included in the original risk assessment. A cobot is purchased and risk-assessed bare. A gripper, suction tool, weld torch, dispenser, or deburring head is later mounted by maintenance or by a system integrator. The new tool changes mass, geometry, sharp edges, payload, and contact characteristics. The original risk assessment is no longer valid. We find this on roughly 70 percent of fenceless cobot audits in West Michigan plants in 2026.
Related reading: Robot Cell LOTO: Aligning ANSI R15.06, ISO 10218, and OSHA 1910.147, Robotics Gap Analysis Service, Annual LOTO Audit.