The remote gripper's fingertip force sensor reports contact force F = kΒ·(g β c), where g is how far the gripper has closed, c is the closing amount at first contact, and k is the object's stiffness. With force feedback on, that force is rendered back into the lever as resistance β the handle visibly stalls short of your commanded position once F approaches your push capacity, so you naturally stop squeezing at a safe force for whatever stiffness you're touching. With feedback off, the lever tracks your hand exactly with zero resistance: you must judge grip purely from how far the fingers look closed on screen, which looks nearly identical for a rigid block and a fragile egg.
F = k Β· max(0, g β c)
resisted: |Ξg| β€ v_maxΒ·dtΒ·(1 β F/P)
blind: g = pointer position (no cap)
Real-world relevance: this is exactly why surgical and space teleoperation systems invest heavily in force-reflecting haptic controllers β a millisecond-scale force channel back to the operator's hand prevents both crushing delicate tissue and dropping heavy tools, something a video feed alone cannot convey.