# Robot gripper design: grip force, pads, vacuum cups and testing

> Design a robot gripper that holds the part: four gripper families, the grip-force formula with a worked example, pad friction, vacuum cup sizing and test steps.

Canonical: https://roboticsengineering.in/robot-gripper-design-basics/ | Published: 2026-10-27 | Updated: 2026-10-28 | Author: Jignesh Panchal

**A robot arm is only as useful as its gripper, and the gripper is usually the part that gets built last and tested least.** A parallel jaw with the wrong pad material will drop a smooth bottle; a vacuum cup sized by guesswork will fail the moment the surface is dusty. This guide covers the four common gripper types, the force calculation, and a build you can finish in an evening.

## The four gripper families

| Type | How it holds | Best for | Fails at |
| --- | --- | --- | --- |
| Parallel jaw | Two jaws close linearly | Boxes, blocks, flat parts, repeatable pick and place | Fragile or irregular objects, very wide size range |
| Angular jaw | Jaws pivot like a pinch | Round and irregular objects, wider opening | Parallel faces; grip force varies with opening |
| Vacuum | Suction cup seals against a surface | Flat, smooth, non-porous panels, cardboard, glass | Porous, dusty, textured or curved surfaces |
| Magnetic | Permanent or electromagnet | Ferrous metal parts, scrap, sheet | Aluminium, plastic, stainless steel |

Soft grippers made from silicone or flexible fins are a fifth option. They conform to irregular produce and fragile parts, but they are harder to model, harder to clean and usually need a pneumatic supply. For a first robot, parallel jaw is the right default.

## Calculate the grip force before designing the jaws

A parallel jaw holds a part by friction. For two jaws, the minimum grip force is:

```
F_grip = (m × g × SF) / (2 × μ)

m  = mass of the object (kg)
g  = 9.81 m/s²
SF = safety factor, 2 for a gentle pick, 3–4 for fast motion
μ  = friction coefficient between pad and object
```

For a 0.5 kg block with rubber pads on plastic (μ ≈ 0.4) and a safety factor of 2, the grip force is (0.5 × 9.81 × 2) / (2 × 0.4) = 12.3 N. That is about 1.25 kg of force, which an SG90 cannot deliver comfortably but an MG996R can.

| Pad material | Friction on plastic | Friction on metal | Notes |
| --- | --- | --- | --- |
| Silicone rubber | 0.6–0.9 | 0.5–0.8 | Best general choice, leaves no residue |
| Soft foam | 0.5–0.7 | 0.4–0.6 | Conforms, but wears and compresses |
| Hard PLA | 0.2–0.4 | 0.2–0.3 | Only for parts with a mechanical stop |
| Sandpaper | 0.7–0.9 | 0.6–0.8 | Grips well, scratches the part |

The friction coefficient is the number people guess and get wrong. A glossy PET bottle with hard PLA jaws has a coefficient near 0.2 and will squirt out of the gripper no matter how hard the servo pushes. Silicone pads solve it for a few rupees.

## Servo torque to grip force

A servo-driven parallel jaw converts torque into force through the jaw arm. Ignoring linkage losses, force is torque divided by the distance from the pivot to the contact point. An MG996R rated at 9.4 kg·cm (0.92 N·m) at 6 V, acting at 30 mm, produces about 30 N at the jaw — more than the 12.3 N the example needs. An SG90 at 1.8 kg·cm (0.18 N·m) at the same distance produces about 6 N, which is marginal.

Two practical corrections. A rack-and-pinion or four-bar linkage changes the force curve as the jaws close, usually reducing force at the fully open position. And the servo’s rated torque is at 6 V, while a 5 V supply through a long wire may deliver 20% less. Build in margin, and use a force-limiting grip: command the servo to a position, and stop when the current rises or after a fixed time rather than stalling it continuously.

## Vacuum cups: area and pressure

A suction cup holds a load through the pressure difference across its area:

```
F = ΔP × A

A 30 mm cup has area π × 0.015² = 7.07 cm²
At −40 kPa (0.4 bar below atmosphere): F = 40,000 × 0.000707 = 28.3 N
```

That is the theoretical force with a perfect seal on a clean, flat surface. Use 50% of it as the working limit for horizontal lifts, and much less for vertical surfaces where shear and peeling dominate. A dusty cardboard box may leak so badly that the cup never reaches the rated vacuum at all.

- **Flat and smooth:** a standard bellows or flat cup works.
- **Curved:** use a bellows cup that can conform, and expect lower force.
- **Porous:** cardboard and foam need a high-flow pump and a larger cup, or a different gripper entirely.
- **Dusty:** add a filter and plan on cleaning cups daily.
- **Hot parts:** silicone cups resist heat; nitrile and rubber may not.

## Design rules that save rebuilds

- **Make the jaws parallel at contact,** not at rest. A four-bar linkage keeps the jaw faces parallel through the stroke.
- **Add a mechanical stop** so the jaws cannot crush the part or stall the servo.
- **Design for the smallest and largest object** in the same stroke, and check that the linkage does not go over-centre at either end.
- **Put the part against a backstop** so its position is repeatable, not just gripped.
- **Keep the centre of mass close to the wrist.** A long gripper multiplies the torque the wrist servo must hold.
- **Print jaws with the layers running across the load,** not along it. PLA splits between layers under a prying load.
- **Add a compliant pad or a spring** so a small misalignment does not jam the gripper.

## A ₹1,500 parallel gripper build

| Part | Spec | Indicative price |
| --- | --- | --- |
| Servo | MG996R, metal gears | ₹350–₹550 |
| Jaws | 3D-printed PETG or PLA, 3 mm walls | ₹100–₹200 of filament |
| Pads | 2 mm silicone sheet, cut to jaw face | ₹50–₹150 |
| Linkage | M3 screws, nuts, two short pushrods | ₹100–₹200 |
| Servo bracket | 3D-printed or aluminium | ₹100–₹250 |
| Controller | Arduino Uno or PCA9685 | ₹400–₹700 |

The mechanism is simple: the servo horn drives one jaw, and a linkage mirrors the motion to the second jaw so the gripper self-centres. The build below assumes the servo is positioned so 0° is fully open and 90° is closed on a 40 mm object. Calibrate the actual angles on your mechanism; every linkage is slightly different.

```
#include <Servo.h>

Servo gripper;
const int OPEN = 20;      // degrees, measured on your build
const int CLOSED = 95;
const int HOLD_MS = 400;

void setup() {
  gripper.attach(9);
  gripper.write(OPEN);
  delay(500);
}

void grip() {
  gripper.write(CLOSED);
  delay(HOLD_MS);
  // Do not leave a stalled servo powered indefinitely on a heavy object.
}

void release() {
  gripper.write(OPEN);
  delay(HOLD_MS);
}
```

For a stronger and more repeatable grip, replace the servo with a small stepper and a lead screw, or add a force sensor to the jaw. The [pick-and-place arm project](https://roboticsengineering.in/pick-and-place-robot-arm-arduino/) shows the gripper in a full pick cycle, and the [servo guide](https://roboticsengineering.in/servo-motor-with-arduino/) covers power and wiring.

## Testing before it goes on the robot

- **Grip the heaviest part at the worst orientation** and shake it gently by hand. If it slips, the pad or force is wrong.
- **Cycle 50 times** and check screw loosening, jaw wear and servo temperature.
- **Test a misaligned part:** place it 5 mm off-centre and confirm the gripper still captures it.
- **Test the failure mode:** cut power mid-grip. Does the part fall, or does the mechanism hold it? Either is acceptable if you know which.
- **Check pinch points** and keep fingers out of the jaw path during testing.

## Common questions

**How much grip force do I need?**

Use F = (m × g × safety factor) / (2 × friction coefficient). A 0.5 kg part with rubber pads needs about 12 N with a safety factor of 2. Fast moves need a higher factor.

**Can an SG90 servo drive a gripper?**

Yes for small, light parts with good pads. It produces about 1.8 kg·cm, so the jaw arm must be short and the object light. An MG996R is the practical minimum for a general-purpose gripper.

**What is the best material for gripper pads?**

Silicone rubber, around 2 mm thick. It has high friction on plastic and metal, conforms slightly, and does not leave residue.

**Why does my vacuum gripper drop boxes?**

Cardboard is porous and dusty, so the cup leaks and never reaches full vacuum. Use a higher-flow pump, a larger bellows cup, or switch to a mechanical gripper.

**Should the gripper be parallel or angular?**

Parallel for boxes and flat parts, angular for round or irregular objects. Parallel jaws give more repeatable positioning because the part stays centred.

Related: [pick-and-place robot arm](https://roboticsengineering.in/pick-and-place-robot-arm-arduino/), [robot arm kinematics](https://roboticsengineering.in/robot-arm-kinematics-basics/), [PCA9685 servo driver](https://roboticsengineering.in/components/pca9685-servo-driver/), [MG996R servo guide](https://roboticsengineering.in/components/mg996r-servo/).

The cell budget usually includes the gripper: [industrial robot cost in India](https://roboticsengineering.in/industrial-robot-cost-india/) separates hardware from integration.

See also: [cobot welding cells](https://roboticsengineering.in/cobot-welding-small-business-india/), where the fixture matters more than the gripper.

Most SCARA projects live or die on the end effector, so read [gripper design basics](https://roboticsengineering.in/scara-robot-guide/) before specifying a vacuum cup or jaw.

The gripper is often the limiting part of a delta cell, which is why [gripper design basics](https://roboticsengineering.in/delta-robot-guide/) is worth reading before you enquire.

Grippers that slide instead of pivot often use a [linear actuator](https://roboticsengineering.in/linear-actuator-guide-robotics/); the sizing rules are different from rotary motors.

Grippers often use pneumatics while the arm uses electric drives; see [actuator type comparison](https://roboticsengineering.in/hydraulic-pneumatic-electric-actuators/) for the trade-offs.

The gripper decides whether a palletizing cell keeps up; see [palletizing gripper and cell costs](https://roboticsengineering.in/palletizing-robot-cost-india/).

Gripper fingers are usually driven through a small gearbox or linkage; see [planetary vs spur gearboxes](https://roboticsengineering.in/planetary-vs-spur-gearbox-robot/) for the trade-off.

The gripper mass counts against [the robot’s usable payload](https://roboticsengineering.in/industrial-robot-payload-calculation/).

Compare the full range in [robot end effector types](https://roboticsengineering.in/robot-end-effector-types-guide/) before designing fingers.

Once the design is settled, prices and options are compared in the [robot gripper buying guide](https://roboticsengineering.in/robot-gripper-buying-guide-india/).

For flat, non porous parts, [vacuum gripper sizing](https://roboticsengineering.in/vacuum-gripper-design-guide/) usually beats a mechanical jaw design.

Gripper design decisions depend on whether the robot can feel the grip, covered in [force torque sensing](https://roboticsengineering.in/robot-force-torque-sensor-guide/).

Air driven grippers are simple, but [cylinder sizing](https://roboticsengineering.in/pneumatic-cylinder-sizing-robots/) decides whether they hold under load.

Food contact changes gripper material and cleanability, as the [food industry robot guide](https://roboticsengineering.in/food-industry-robots-india/) sets out.

## Sources and further reading

- [Festo handling catalogue](https://www.festo.com/gb/en/p/electronic-catalogue/) — gripper sizing, vacuum cup force and safety factors.
- [Schunk](https://www.schunk.com/) — parallel and angular gripper specifications for comparison.
- [Robot gripper](https://en.wikipedia.org/wiki/Robot_gripper) — overview of jaw, vacuum and soft gripper designs.

Last reviewed: October 2026. Force figures assume clean, dry contact surfaces; test on the real part before relying on a grip. Prices are Indian retail ranges. Not bench-tested by this site.