makermods / shop / metal arm
MakerMods Metal Arm, six-axis lightweight robotic arm
METAL ARM · CNC ALUMINUM
612.5 mm REACH · 4.2 kg 6 DOF · ±0.1 mm
[ LIGHTWEIGHT 6-AXIS ROBOTIC ARM · EMBODIED AI ]

MakerMods
Metal Arm.

A lightweight, all-CNC-aluminum six-axis manipulator built for embodied AI, research, and education. 3 kg payload, ±0.1 mm repeatability, open SDK, ready for ROS, MoveIt, LeRobot, and VLA training out of the box.

$2,499 from $99 deposit to pre-order · fully assembled ~1 week lead
[ PRE-ORDER → ]
mit license · open sdk · ros1 / ros2
[ AT A GLANCE · 02 ]

From sim to real hardware, without seams.

High-precision motion control, low-power design, and an open hardware-software stack, engineered for research labs, university teaching, and embodied-AI startups.

MakerMods Metal Arm, six-axis profile
01 · ARM
6 DOF±0.1 mm
2-in-1 end-effector, gripper plus teacher
02 · END-EFFECTOR
swap in seconds0–95 mm stroke
Operator guiding the arm under gravity compensation
03 · GRAVITY COMP
out of the boxteleop by hand
[ HW · 01 ]

All-CNC metal build

Aluminum-alloy housing, durable, lightweight, and rigid enough for hands-on teaching and field deployments.

[ HW · 02 ]

Modular end interface

Quick-swap between gripper, teacher, and 2-in-1 end-effector. Power + CAN through a single XT30 (2+2) connector.

[ HW · 03 ]

±0.1 mm repeatability

Precision suitable for research-grade pick-and-place, contact-rich manipulation, and fine assembly tasks.

[ HW · 04 ]

Ready out of the box

Calibrated, packed, and shipped complete, unbox, plug in, and run your first experiment in minutes.

[ END-OF-ARM TOOLING · 03 ]

Three swappable end-effectors.

Pick the tool that fits your workflow, gripper, guide for teleoperation, or both at once. All three share the same XT30 power + CAN interface and swap in seconds.

Gripper

Gripper

weight670 g stroke0–95 mm repeat±0.1 mm interfaceXT30 2+2
Leader

Leader

weight670 g stroke0–95 mm repeat±0.1 mm interfaceXT30 2+2
2-in-1 end-effector

2-in-1

weight670 g stroke0–95 mm repeat±0.1 mm interfaceXT30 2+2
[ SOFTWARE & CONTROL · 04 ]

Joint, end-pose, teach, and trajectory.

A stable IK solver, adaptive trajectory tracking, gravity compensation, and a full simulation pipeline, all driven by an open-source Python and C++ SDK.

Operator demonstrating gravity-compensated teaching
[ 4.1 · GRAVITY COMPENSATION ]

Teach by hand.

Lightweight gravity compensation lets an operator guide the arm directly through a motion, ideal for hands-on data collection and rapid demo capture.

Adaptive trajectory tracking demonstration
[ 4.2 · PRECISE CONTROL ]

Joint + end-pose control.

Precise control: supports joint and end-pose control, with a built-in stable IK solver.

URDF visualization of the MakerMods Metal Arm
[ 4.3 · URDF SUPPLIED ]

Drop into your stack.

Ready-to-use URDF for the arm and every end-effector configuration, for visualization, planning, and downstream development without rebuilding kinematics.

MoveIt and Gazebo co-simulation
[ 4.4 · MOVEIT + GAZEBO ]

Sim today, hardware tomorrow.

Pre-built MoveIt and Gazebo integration with real-hardware loopback. Develop and validate in sim, then deploy on the physical arm without rewriting glue.

[ CONTROL MODES · 05 ]

Two control modes.

Direct control or teleoperation, switch modes by loading a different SDK config. The full data-collection codebase is open source.

Direct control, operator manipulates the arm directly
5.1 · DIRECT CONTROL

Hands-on the arm.

arm × 1  +  2-in-1 × 1

Single-handed grip on the 2-in-1 end-effector. Guide the arm directly to capture motion.

Teleoperation, leader controls a follower arm
5.2 · TELEOPERATION

Leader + follower.

arm × 2  +  Leader × 1  +  Gripper × 1

Operator holds the leader to remote-control the follower arm at the gripper end.

[ WHAT'S IN THE BOX · 06 ]

Inventoried, calibrated, packed.

Complete packaging, everything you need to stand up a teleop, capture, or sim experiment the day it lands on your bench.

01
Robotic arm
6-DOF CNC-aluminum arm, factory-calibrated, ready to mount.
×1
02
2-in-1 end-effector
Combined gripper + leader unit. Default end-effector.
×1
03
Gripper
Parallel-jaw gripper. 0–95 mm stroke.
×1
04
Leader (handheld)
Leader-side teleop unit with handle.
×1
05
CAN-to-USB adapter
PC-side bus interface. Driverless on Ubuntu 20.04 / 22.04.
×1
06
Switching power supply
24 V DC supply with regional plug adapters.
×1
07
Metal stabilizer plate
Bench mounting plate, drilled for the 90×90 mm M5×4 base pattern.
×1
08
G-clamp + toolkit
Clamp the stabilizer plate to any bench. Hex keys + power button included.
kit
[ ONE PLATE · 4 SCREWS ]

Pre-built and factory-calibrated.

Ships ~95% assembled. Clamp the stabilizer plate to your bench, mount the arm with 4 × M5, plug in power + CAN-to-USB. You're sending joint commands within minutes.

  • Joints homed and calibrated at the factory
  • SDK + URDF + ROS drivers preinstalled-ready
  • 4-screw bench mount, no welding, no fabrication
  • Foam-packed for international freight
[ SPECS · 07 ]

The numbers. The specs.

metal arm · built by makermods spec_sheet.txt
body weight4.2 kg9.3 lb
rated payload3 kg6.6 lb
degrees of freedom6
working radius612.5 mm24.1"
repeatability±0.1 mm
base mount90 × 90 mm · M5 × 4
supply voltage24 V DC
materialCNC aluminum alloy + resin
communicationCAN
external interfacePower + CAN · XT30 2+2
control modesTrajectory tracking · Teach · API
joint range · J1 / J6−165° to 165°  ·  −150° to 150°
max joint speed180 °/s (J1–J3) · 220 °/s (J4–J6)
ros supportROS1 Noetic · ROS2 Humble
sdkPython + C++ · open source · continuously maintained
simulationMoveIt · Gazebo (real-hardware loopback)
vla supportALOHA ACT · pi0 / pi0.5 (training + inference)
licenseMIT (SDK + firmware)
[ FAQ · 08 ]

Things people ask before buying.

Yes, ships ~95% assembled and factory-calibrated. On your end: clamp the metal stabilizer plate to a bench, fasten the arm to the plate with 4 × M5, and plug in power + the CAN-to-USB adapter. You're sending joint commands within minutes.
Any PC running Ubuntu 20.04 or 22.04. The CAN-to-USB adapter is driverless on both. ROS1 Noetic and ROS2 Humble are supported; the Python and C++ SDKs both ship with ROS1 and ROS2 drivers.
Yes. We provide full open-source reproductions of ALOHA ACT and pi0 / pi0.5 with training and inference pipelines plus step-by-step tutorials. Additional VLA reproductions are rolling out.
Yes, MIT license, hosted on github.com/Maker-Mods. Python + C++ SDKs, ROS1/ROS2 drivers, the data-collection codebase, URDF files, and the VLA reproductions are all open and continuously maintained.
Yes, we sell a dedicated Education Package bundled with a teaching curriculum and lab materials. Email isaac@makermods.ai with your program and class size.
Yes, single-arm and bimanual configurations are both first-class. The data-collection codebase ships with four capture configs out of the box: single-arm direct, single-arm teleop, dual-arm direct, and dual leader / dual follower teleop.