A robotic arm gripper is the tool at the very tip of the arm that actually does the work, and it is where most DIY builds either succeed or fail. You can perfect every joint and still end up with an arm that drops everything, because grip is a problem of geometry, force, and material, not just motor power.
What an end effector actually is
The end effector is whatever you bolt to the last joint of the arm to interact with the world. A gripper is the most common end effector, but the category also includes vacuum cups, magnets, tool changers and specialised tools — see end effector types for the full rundown. For most hobby builds following a step-by-step robotic arm build guide, the end effector is a two-finger gripper driven by a single servo.
The gripper has to do three things at once: close around an object, generate enough force to hold it against gravity and slip, and release cleanly. A surprising number of designs nail the first and fail the other two.
Gripper types compared
There is no universal best gripper. The right choice depends on what you are picking up: rigid blocks, round bottles, eggs, or flat sheets. Here is how the main families stack up.
| Gripper type | How it works | Best for | Pros | Cons |
|---|---|---|---|---|
| Parallel-jaw | Two jaws stay parallel and translate together | Boxes, blocks, machine parts | Even contact, predictable, secure | Needs gear/rack mechanism, more parts |
| Claw / pincer linkage | Jaws pivot inward on a four-bar linkage | General hobby pick-and-place | Simplest single-servo build, cheap | Contact point shifts as jaws close |
| Soft / compliant (TPU) | Flexible fingers bend around the object | Eggs, fruit, irregular shapes | Forgiving, no crushing, self-aligning | Low max force, slower, wears out |
| Vacuum / suction | Suction cup + pump or syringe | Flat, smooth, non-porous parts | Single contact point, no crushing | Needs vacuum source, fails on porous/rough items |
| Hook / specialized | Fixed or single-DOF custom shape | Repetitive, known objects | Dead simple, very strong | Only works for one object class |
All of these are servo-driven in the table above, but a jaw that needs long, high-force travel is sometimes better served by a linear actuator than a servo-and-linkage — see that guide for when the swap is worth it.
Tip: If you are building your first arm, start with a pincer linkage — the classic robot claw. It is the easiest to print and tune. You can graduate to parallel-jaw once you understand the force trade-offs.
How one servo drives a linkage gripper
The elegant part of most DIY grippers is that a single servo opens and closes both jaws symmetrically. Two mechanisms dominate.
Meshed gears. Each jaw has a gear sector at its pivot. The servo horn drives one jaw directly; the second jaw is geared to the first, so they mirror each other. This is compact and gives a clean parallel-ish motion, but printed gear teeth need decent resolution or they bind.
Four-bar linkage. The servo rotates a crank, which pushes a coupler link, which swings the finger. Duplicate it mirrored for the second finger off the same servo horn. Four-bar linkages are tolerant of loose printed tolerances because the load passes through pin joints rather than meshing teeth, which is why they are the most common hobby choice.
In both cases the linkage acts as a lever. A longer servo-side arm and a shorter finger-side arm trades range of motion for clamping force. A 3:1 or 4:1 ratio is a good starting point.
Matching grip force to servo torque
This is the calculation people skip and then wonder why the gripper drops things. Grip force at the fingertip is the servo torque divided by the effective lever arm, multiplied by the linkage advantage.
A standard hobby servo choice for robotic arms breaks down roughly like this:
| Servo | Stall torque | Realistic gripper duty | Typical fingertip force |
|---|---|---|---|
| SG90 (plastic gear) | ~1.8 kg·cm | Objects under ~150 g | Light, hold gently |
| MG90S (metal gear) | ~2.2 kg·cm | Up to ~250 g | Moderate, more durable |
| MG996R (metal gear) | ~10 kg·cm | Up to ~500 g, slippery items | Strong, can crush soft items |
Use stall torque only as a ceiling; run the servo at 30-50% of stall for continuous holding so it does not overheat or buzz. If you need a firm grip on heavier parts, do not just buy a bigger servo, increase the linkage ratio first. The same MG996R used on an Arduino-controlled robotic arm can roughly double its holding force just by changing the link lengths.
Tip: Add a strip of TPU, silicone, or even bicycle inner tube to the jaw faces. The grip force you can hold scales with friction, and a rubber pad often beats buying a stronger servo.
3D-printing tips for gripper parts
Most of a DIY gripper is printed, and a few choices make the difference between crisp action and a sloppy, binding mechanism. These apply whether you are remixing a design or starting from a full 3D-printed robotic arm project.
- Orient pivots vertically so the holes print round and the layer lines run around the pin, not across it. Holes printed flat come out egg-shaped.
- Print pin joints with 0.2-0.3 mm clearance. Tight pins seize; loose pins wobble. Dial it in with a tolerance test print.
- Use PETG or PLA for the structural links at 30-40% infill. The crank and coupler take the load and need to stay stiff.
- Add a fillet at the servo-horn mount. That joint sees the highest stress and is the first thing to snap.
- Reinforce screw bosses with at least 3 perimeters so self-tapping screws do not split the plastic.
Soft and compliant grippers for fragile objects
When the payload is an egg, a tomato, or a 3D-printed part you do not want scratched, a compliant gripper beats a rigid one. The whole finger is printed in TPU (around 95A shore) so it bends around the object and spreads contact over a large area instead of concentrating it on two points.
The simplest version is a “fin-ray” finger: a triangular rib structure that curls inward when it touches something, wrapping the object passively with no extra actuators. One servo still drives the pair. Because TPU is forgiving, you can drive it harder without crushing, but expect lower absolute force and slower cycles. These designs are a popular entry in most beginner robotic arm projects because they look impressive and hide a lot of tolerance sins.
Mounting the gripper as the last joint
The gripper attaches to the wrist, which is the final joint of the arm. Two common setups:
- Fixed mount: the gripper bolts straight to the forearm. Simple, rigid, but the jaws always face one direction.
- Wrist-rotate mount: a small servo between forearm and gripper rotates the whole end effector, letting you orient the grip. This is what most pre-built robotic arm kits ship with.
Keep the gripper light. Every gram at the tip is the worst-case load for every joint below it, multiplied by the arm’s reach. Route the servo wire along the arm with a little slack and a strain-relief loop near the gripper so repeated motion does not fatigue the solder joints. Test the full open-close cycle with the arm fully extended before you trust it with anything fragile.
Frequently asked questions
What is a robot claw and how is it different from a gripper?
In hobby robotics the terms are used interchangeably — a robot claw is simply a gripper whose jaws pivot inward like a pincer rather than sliding in parallel. Claw designs are the most common DIY choice because a single servo can drive both jaws through a four-bar linkage, making them cheap to print and easy to tune. Parallel-jaw grippers hold flat-sided objects more securely but need a rack or gear mechanism.
How do you make a robot claw?
Print or cut two jaws and a four-bar linkage, mount one servo to drive the crank, and mirror the linkage so both jaws close together. Keep pin-joint clearance around 0.2-0.3 mm so the mechanism moves freely without wobbling, use a 3:1 or 4:1 linkage ratio to trade travel for clamping force, and add rubber or TPU pads to the jaw faces — friction adds more holding power than a bigger servo does.
How much torque does a robotic arm gripper servo need?
For light objects under 200 g an SG90 (1.5-2.5 kg·cm) is enough. For a 2-4 cm grip on heavier or slippery items, use an MG996R (9-11 kg·cm) with metal gears. Grip force depends on linkage ratio, not just raw torque, so a 4:1 linkage roughly doubles holding force versus a direct drive.
What is the best material for 3D-printing gripper jaws?
Print the structural linkage in PLA or PETG for stiffness, then add TPU pads or a fully TPU compliant finger for grip. TPU (around 95A shore) conforms to odd shapes and protects fragile objects. PLA is fine indoors; PETG handles heat and impact better.
Can one servo drive a robotic arm gripper?
Yes. A single servo is the standard approach. A four-bar linkage or a pair of meshed gears converts one servo's rotation into two symmetric jaws that close together, so you only spend one PWM channel on the end effector.