Collaborative robots that share workspace with people without a fence rely on Power and Force Limiting (PFL), the approach formalised in ISO/TS 15066. Two contact scenarios are evaluated separately, because the human body tolerates them very differently:
Transient (impact) contact — a moving mass hits tissue and rebounds:
μ = (m_H · m_R) / (m_H + m_R) reduced (effective) mass
1/k = 1/k_R + 1/k_H contact stiffness in series
F_peak = v · √(k · μ) spring-mass impact force
Quasi-static (clamping) contact — the robot keeps pushing, e.g. pinning
a limb against a fixture. Force/torque sensing must cap the applied
force at a configured limit before it ever reaches biomechanical harm:
F_peak = P_set (the robot's own configured force-limit setting)
Every body region has a different tolerance: soft, low-mass regions like the hand can absorb more localized force before injury than the skull or neck, and the standard allows a higher force for a brief transient impact than for sustained clamping of the same region — the numbers this simulator uses follow that same structure and are representative teaching values, not a certified compliance table.
- Region — sets the body's effective mass mH, contact stiffness kH, and its transient/quasi-static force limits.
- Speed & reflected mass — set v and mR, which only matter for a transient impact; a quasi-static clamp force does not depend on how fast the arm arrived.
- End-effector stiffness — a softer, padded gripper lowers the combined contact stiffness k and therefore the peak impact force, exactly why real cobots use compliant covers.
- Pset — the maximum force the robot's controller allows itself to apply once torque/force sensing detects sustained contact.
Real-world relevance: this two-scenario force budget is why cobots such as Universal Robots' UR series or KUKA's LBR iiwa can legally slow down (not stop) near a person, and why every certified deployment fixes a maximum speed and payload for each robot pose before it is allowed to run without a cage.