Research platform

MiPA for Research

A mobile dual-arm robot with open interfaces for research in mobile manipulation, human-robot interaction and Physical AI.

2 × 7 DoFlightweight arms
3.0 kgpayload per arm
1 kHzlow-level arm control
Jetson ThorNVIDIA onboard compute
ROS 2 + Pythontwo control levels
4 simulatorsconfigured for MiPA

Research

One robot for manipulation, interaction and AI research

Open down to the joint

Control MiPA natively through ROS 2, or use the Python SDK for higher-level commands. The arm controllers take position, velocity and torque commands at 1 kHz.

Two arms, two hands, one mobile base.

Two 7-DoF arms on a linear lift axis and an omnidirectional base. End effectors are changeable, so you can set MiPA up for your own grasping and manipulation experiments.

Built for Physical AI

NVIDIA Jetson Thor runs your models on board. Record arm joints at 1 kHz and video at 30 Hz, export them as ROS bags and feed them into your training pipeline.

Ready in simulation

Isaac Sim, Isaac Lab, MuJoCo and Gazebo come configured for MiPA. Your team can start before the robot arrives.

How you work with MiPA

Two ways to control

Pick the level your project needs. Use ROS 2 for full flexibility down to the arm controllers, or the Python SDK for navigation, arm motion and MoveIt in a few lines of code.

Overhead view of MiPA with a planned navigation path overlaid

Built for interaction studies

MiPA has a touch display, a microphone array and a speaker, so you can study multimodal human-robot interaction on the same robot you use for manipulation.

Close-up of the MiPA touch display

Teleoperation and demonstrations

Control MiPA remotely with common teleoperation devices and record the demonstrations you need for training.

Researcher teleoperating MiPA with a VR headset and controllers

Research

Everything a research lab needs, in one robot

MiPA mobile dual-arm robot with its main components marked

Technical specifications

Dimensions
789 × 606 × 1457 mm
System
Mobile dual-arm system with linear axis
Arms
2 × 7-DoF lightweight arms, up to 3.0 kg payload per arm
Motion endurance
4 to 8 h, depending on the load

Specifications are subject to change.

Test your ideas in simulation first

01

Isaac Sim

Generate synthetic data. Comes configured with MoveIt and Nav2.

02

Isaac Lab

Train MiPA policies with reinforcement learning.

03

MuJoCo

Run physics simulation with a ready robot model and configuration.

04

Gazebo

Simulate inside your existing ROS environment.

MiPA simulated in NVIDIA Isaac Sim

From data to deployed skill

MiPA training in a NEURA Gym kitchen setup

NEURA Gym

Train and test skills on real NEURA robots in NEURA facilities, or collect data with your MiPA on site and use the NEURA Gym data pipelines.

Learn more about NEURA Gym →
Digital twin of MiPA in a kitchen scene in Neuraverse

Neuraverse

Build, simulate and deploy on one platform, with a digital twin, visual flow-based programming, Neuron Suite, Creator Hub and Marketplace.

Early access →

NEURA Research Programme

NEURA Research Programme 2026/27

The 2026/27 call closed on 30 October 2026 with more than 170 proposals. The NEURA Research Board selects projects by 20 November 2026, and the first projects start from 1 December 2026.

Up to €500,000per project
Up to €10 millionexpected for the call
12 to 18 monthsproject duration
3 segmentshealth care, retail, hospitality

Where we are

30 Oct 2026Call closed
By 20 Nov 2026Review and selection by the NEURA Research Board
From 1 Dec 2026Start of the first funded projects

Project start is subject to a separate cooperation agreement with NEURA.

Funded projects

Funded projects will be presented here after the selection.

Next call

No further call has been announced yet. For inquiries contact us at [email protected]

The NEURA Research Programme is a private-sector funding initiative of NEURA Robotics GmbH. It is not a public grant and not a tender, and it does not constitute an offer.

Frequently asked questions

What is MiPA?

MiPA is NEURA's mobile dual-arm robot with an open software interface. Research teams use it for mobile manipulation, human-robot interaction and Physical AI.

Can I use my own software?

Yes. MiPA is ROS 2 native. You can send position, velocity and torque commands straight to the arm controllers.

Which simulators are supported?

Isaac Sim (with MoveIt and Nav2), Isaac Lab, MuJoCo and Gazebo.

Can I use my own end effectors?

Yes. End effectors are changeable, and the flange provides 24 V and Single Pair Ethernet.

How do I get a MiPA?

Tell us about your project in the form below and our team will get back to you with options.

Can I still apply for the Research Programme?

No. The 2026/27 call closed on 30 October 2026. Leave your details in the form below to learn more about future programmes.

Tell us about your research

Tell us about your team and what you plan to research with MiPA. We will get back to you.

Scroll to Top