Dobot Dobot X-Trainer (Dobot X-Trainer)

A robot learns a task only as well as the person who demonstrates it, and most people tire quickly of demonstrating. The Dobot Dobot X-Trainer (Dobot X-Trainer) is built around that human limit. An operator holds two light master hands, two Nova2 arms follow every move, and the whole session is recorded as training data. The system is a teaching station first and a robot second.

The product is available through this page, by quotation.

Background

Dobot and the road to embodied AI

Dobot is a robotics maker best known for desktop and collaborative arms, and the Nova2 cobot at the heart of this system comes from that lineage. In April 2024 the company released the X-trainer AI Data Collection and Training Robotics System to support embodied AI agents. It followed the same idea that has pushed many labs toward teleoperation: the quickest way to teach a machine a physical skill is to let a person show it.

Other Dobot platforms, from humanoids to legged robots, are gathered in the Dobot category. The X-Trainer sits at a different point in that range. It is not the end product that walks or works in the field. It is the equipment that produces the demonstrations such machines learn from.

How the system is positioned

The DOBOT X-Trainer is an advanced AI data collection and training robotic system designed to enable AGI research and practical AI projects. Its output is data in formats suitable for Mobile ALOHA training, so teams working with that style of imitation learning can use it without building their own conversion tools.

The hardware is a matched pair: two teleoperated master hands for the operator and two Dobot Nova2 slave arms, each with its own end gripper and end camera. A removable mounting platform carries a global camera. The sources also name a DOBOT X-Trainer (Lightweight Base) version, so you can ask about the base that suits your bench or lab.

Technical Specifications

Specifications as published by the manufacturer; confirmed against the datasheet at quotation.

The manufacturer has not published the dimensions, the weight, the power supply, the power consumption, the camera resolution, the gripper type and stroke, the operating temperature or the host computer requirements; they are confirmed at quotation.

What makes the X-Trainer different from a standard cobot setup?

A standard collaborative arm is programmed with waypoints or a teach pendant. The X-Trainer is taught by demonstration. The operator moves a master hand, and the matching Nova2 arm copies the movement while cameras and joint data are recorded. The result is a record of how a person handles a task, with all the small corrections included, which a waypoint program does not capture.

Several details serve that purpose. The 25Hz end-to-end motion interface improves response speed by 150% compared with similar products and is more stable. That matters because lag between the operator's hand and the arm makes demonstrations clumsy. The master hands carry a load of under 400g and use 3D printed monolithic joints, which means fewer parts and a lighter feel over a long session. Each master hand has an unlock button and a record button. The unlock button locks or unlocks the master hand pose quickly to reduce fatigue, and the record button controls data capture without the operator reaching for a keyboard.

Training time is also shorter. Routine scenario training, such as dish-washing or folding clothes, can be completed in just 2H, a reduction of 70%.

How does the X-Trainer protect people and equipment?

Teleoperation puts a person very close to moving machinery, so the safety design is part of the product. The system auto de-syncs the master and slave arms on power loss or alarms. It detects when a master hand is dropped, and it offers low-speed master-slave sync to prevent collisions. The Nova2 cobot has 5-level collision detection and ISO15066 certification for safe human-robot interaction.

The work-space is padded with flexible EVA foam. It cushions accidental drops of the end-effector and lowers the risk of damage to the gripper and arm. Students and new operators will make those mistakes, and the padding makes it easier to forgive them.

Who is the Dobot X-Trainer for?

University and corporate research groups working on imitation learning and embodied AI are the clearest fit. They need many consistent demonstrations, and they need open access to the data. The open APIs expose master and slave joint angles, gripper control and camera RGB and depth, so a team can connect its own models. The complete SDK covers data collection, model training and autonomous inference, with documentation that guides users through the whole data pipeline.

Teaching labs benefit too. A course on robot learning can take students from recording a demonstration to running a trained policy on one system. Companies exploring service tasks, such as household or hospitality routines, can use it to test whether a task can be learned before committing to a larger platform.

If your goal is a humanoid or mobile platform rather than a data collection station, look at the Dobot Atom Max or the Dobot Atom D. Buyers who are unsure which base of the X-Trainer fits their space can ask about the Lightweight Base version when requesting a quotation.

What is the Dobot X-Trainer used for?

1. Demonstration capture: an operator records dual-arm tasks through the master hands, with the record button controlling the session.
2. Imitation learning research: the data format suits Mobile ALOHA training, so recorded sessions feed straight into model work.
3. Household task training: routine scenarios such as dish-washing or folding clothes can be trained in about 2H.
4. Custom model integration: open APIs for joint angles, gripper control and camera RGB and depth connect your own AI models.
5. Precise experiments: repeat positioning accuracy of ±0.05mm supports sensitive data collection tasks.
6. Teaching and demonstration: the SDK and documentation guide students through collection, training and inference.

Pricing and Availability

We do not list prices for the Dobot Dobot X-Trainer (Dobot X-Trainer). Please request a quotation, tell us about your project and the version you are considering, and we will respond. Delivery, customs and support are arranged together with the quotation. The product and its quotation are available on this page.

Frequently Asked Questions

What does the system include?

It includes two teleoperated master hands, two Dobot Nova2 slave arms with two end grippers and two end cameras, and a removable mounting platform with a global camera.

How accurate is it?

Repeat positioning accuracy is ±0.05mm. This supports precise experimental operations and sensitive data collection tasks.

What reach and payload can I expect?

Each arm reaches 625mm, expandable to 1200mm in dual-arm mode. Payload is 2kg for a single arm and 3kg for dual arms, with a top speed of 1.6m/s.

Can I connect my own AI models?

Yes. Open APIs give access to master and slave joint angles, gripper control and camera RGB and depth. The SDK covers data collection, model training and autonomous inference.

Is it safe for beginners?

The system de-syncs the arms on power loss or alarms, detects master hand drops and offers low-speed master-slave sync. The Nova2 cobot has 5-level collision detection and ISO15066 certification, and the work-space is padded with EVA foam.

Which training data format does it produce?

It generates data formats suitable for Mobile ALOHA training.

Where can I buy the Dobot X-Trainer and what does it cost?

You can buy it through this page by requesting a quotation. We do not publish prices; the cost is given in the quotation, together with delivery, customs and support arrangements.

Summary

The idea behind this system is that the operator is the bottleneck, so the equipment is built to make demonstrating easier. Light master hands, a 25Hz motion interface, record and unlock buttons and padded safeguards all serve that goal. Precise Nova2 arms, an open SDK and Mobile ALOHA-ready data turn those demonstrations into training material. For teams that want to teach physical skills to AI, this page is the starting point.