Real robotic fingertips carry piezoresistive or optical tactile matrices that read normal pressure at hundreds of hertz. Static contact produces a steady pressure signal; the instant an object begins to micro-slip inside the grip, friction briefly drops below the holding threshold and the pressure signal picks up a distinctive high-frequency vibration — the "slip signature". A controller watching for that signature can react in milliseconds, long before the slip becomes visible or the object falls.
- Adaptive mode — starts at the minimum force that should hold the object at rest. When lift-off acceleration or a lateral jolt overcomes static friction, the fingertip sensors detect the resulting micro-slip velocity and the controller bumps the grip force up in small steps until slipping stops — never squeezing harder than necessary.
- Naive fixed-force mode — grips at one constant, deliberately high force from the very start so it never has to react to slip. It survives any jostle, but for a fragile shell that fixed force alone can exceed the object's crush limit before the lift even begins.
- The dilemma — too little force and inertia during acceleration overcomes friction and the object slides free; too much force and the shell fails structurally. Slip detection lets the controller ride the narrow band between the two in real time instead of guessing a single number up front.
This is the same side-view mechanic as the 3D scene, redrawn flat: a pair of fingers squeezing an egg from the side, with the tactile pad color and the force chart telling the same story a real slip-sensing gripper's control loop would see.