grasp
A grasp is the moment a robot hand takes hold of an object, touching it at a few well-chosen spots so that the object can no longer slip, tip, or fall out. Think of how you pick up a coffee mug: your fingers press on a few places, and once they do, the mug moves with your hand instead of sliding around on its own. A robot grasp is the same idea — a small set of contact points that, together, pin the object to the hand and let the robot carry, lift, or place it with confidence.
What makes a grasp good is not how hard the robot squeezes but where it touches and how those touches push back against the world. Every shove, tug, or shake on the object is a force trying to move it; a solid grasp arranges its contacts — and the friction at each one — so that they can balance out whatever the world throws at the object, keeping it locked to the hand. A bad grasp leaves a gap: some twist or slide the contacts cannot resist, and the object wobbles or drops. So a grasp is really a little balance of forces, not just a clamp.
Grasping is harder for robots than it looks because objects come in endless shapes, weights, and surfaces, and the hand often cannot see exactly where everything is. A grasp that works on a smooth box may fail on a slippery bottle or a floppy bag. This is why grasping sits at the heart of robot manipulation: almost every useful job — sorting parcels, loading a dishwasher, assembling a part — begins with the simple-sounding but surprisingly deep act of getting a firm hold.
A two-fingered gripper grasps a wooden block by pressing on its two flat sides; the block stays put even when the arm swings, because the two contacts and their friction balance out the tugs of motion.
A grasp is a few contacts working together to hold an object steady.
In robotics, a grasp usually means the specific set of contact points and hand pose used to hold one object, not the general ability to grasp.