Aloha Mobile AI Robotic Machine Learning Kit
The Mobile AI is a high-performance, mobile all-in-one robotic machine learning workstation built to meet the demand for affordable, capable data collection and model development. With a four-arm configuration, an intuitive touchscreen interface, and a high-performance laptop, it delivers cutting-edge performance right out of the box—accelerating research in automation, human-robot interaction, and advanced manipulation without the need for extensive setup or modification.
SPECIFICATIONS
2X Leader-Follower
WidowX AI Arm Pairs
3X Intel RealSense
D405 Cameras
1X 10" Touchscreen
Frame & Accessories
Without Laptop
Computing capabilities depend on factors such as OS, RAM, storage, and CPU/GPU performance.
$33,695.95
With High-Performance Laptop
Data Collection
Teloperation
On Device Model Training*
Cloud Model Training
Model Evaluation*
$37,845.95
CONFIGURATIONS
Average production time is 2-3 weeks, regardless of shipping rate, service, or speed. Contact sales@trossenrobotics.com for more information.
* The computers offered with our AI hardware are specifically optimized and guaranteed to support model training and/or evaluation for ACT/ACT++ trained models. Performance and compatibility with other machine learning models may vary. We do not guarantee that these computers will meet the requirements for evaluating other current or future machine learning models, as computational demands may differ.
Trossen Robotics reserves the right to substitute computer hardware models, manufacturers, or specifications based on availability, supply chain constraints, or market pricing fluctuations beyond our control. While we strive to provide the highest quality components as advertised, equivalent or superior alternatives may be used to ensure timely order fulfillment.
Substituted hardware will meet or exceed the performance and compatibility requirements necessary for the intended use. Trossen Robotics makes no guarantees regarding the availability of specific brands, model numbers, or configurations and shall not be held liable for variations in components resulting from unforeseen supply chain disruptions.
By purchasing our products, the customer acknowledges and agrees that substitutions may occur and that Trossen Robotics maintains sole discretion in determining appropriate alternative hardware solutions
RESOURCES
From startup guides to expert support, explore the tools, documentation,
and protection that come standard with every Trossen Robotics product.
THE TROSSEN PROMISE
We stand behind our products with an industry-leading commitment to reliability, service,
and long-term support—because we believe performance should be measured in years, not months.
ROBUST 1-YEAR WARRANTY
LIFETIME SUPPORT FOR
TROSSEN PRODUCTS
REPLACEMENT PARTS AVAILABLE
FOR EASY USER REPAIR
COMPARE AI KITS
Solo AI is designed for data collection in the field and for those NOT needing bimanual manipulation.
FAQs
2x Solo AIs CANNOT be converted into a Stationary AI, as we do not sell the frame, cables, and accessories individually.
Stationary AI is designed for those who require bimanual manipulation and a well-controlled lab environment where variables like arm and camera placement are consistent from session to session.
FAQs
Stationary AI can be used as two Solo AIs for data collection in the field.
Stationary AI does NOT come with the table due to the high shipping cost. We recommend a desk with a minimum width of 48" and a minimum depth of 30". The choice of table, including its color, texture, and stability, can have an impact depending on your final use case. Many of our customers aim to implement or fine-tune models like Pi0, Groot, SmolVLA, or Gemini. These models were typically trained on datasets that used a maple-colored tabletop, which is why we recommend the Uline Industrial Packing Table with a maple top.
Here are some key reasons for this recommendation:
-
Visual consistency: Using a similar table helps reduce the domain gap when fine-tuning models trained on standard datasets.
-
Textured surface: The subtle features on the maple surface improve spatial understanding for the model.
-
Color contrast: Light wood tones contrast well with most everyday objects, improving object detection and segmentation.
-
Non-reflective finish: The matte texture minimizes glare and light reflections, leading to cleaner image data.
-
Stability: The table is sturdy, reducing vibrations during demonstrations or evaluation.
Any table with a similar size (approximately 48 inches by 30 inches) and a matte, light wood finish such as maple should work fine. We've observed that using bright or medium-toned wooden surfaces (like maple) helps reduce uncertainty during training by providing good visual contrast and stable surface features. However, other colors, such as black or white, can also be used—as long as you’re aware that you may need to collect more demonstrations and perform additional tuning to achieve comparable results.
If you plan to train models from scratch or operate in a different visual domain, the table color becomes less critical. In that case, collecting more diverse demonstrations can help account for variations in the environment.
Mobile AI is designed for those who require bimanual manipulation in the field, where variables like arm and camera placement are consistent from session to session.
FAQs
Mobile AI can be used as two Solo AIs for data collection in the field.
 |  |  |
|---|---|---|
 |  |  |
 |  |  |
SOFTWARE SUPPORT
OUR AI MISSION
At the core of our AI hardware line is the ethos of the ALOHA Project—A LOw-cost HArdware—bringing advanced robotics to researchers, students, and engineers at an unparalleled price, making cutting-edge ML research more accessible than ever.
