Starling 2 Max for indoor stock control – indoor flight & software integration questions
-
Hello everyone,
My name is Pablo Tarancón, and I work at ITI (Valencia, Spain).
We are currently evaluating the Starling 2 Max drone for a potential purchase and would appreciate feedback from anyone with experience using it (or similar drones) in indoor environments.Our intended use case is indoor stock control, integrating the drone with our own AI model and proprietary software.
I would like to clarify the following points:
Indoor flight conditions
The drone would operate indoors with a ceiling height of approximately 3 meters.
Is this height sufficient for safe and stable flight?
Are there known limitations or risks when flying in such confined indoor spaces?Suitability for our use case
Is the Starling 2 Max suitable for indoor stock control or inventory management applications?
Has anyone used it with custom software or AI models?
Are APIs, SDKs, or technical documentation available to support custom integration?Flight controller and onboard software
Which software or firmware comes pre-installed on the flight controller?
What built-in safety features are available (collision avoidance, emergency stop, geofencing, etc.)?
Does it include stabilization systems specifically designed for indoor flight?
Is a “return to home” or similar failsafe function available for indoor operation?Any insights, experiences, or recommendations would be very welcome.
Thanks in advance for your help!
Best regards,
Pablo -
Hello @Pablo-Tarancón
My two cents as user and developer with Starling 2:
The drone would operate indoors with a ceiling height of approximately 3 meters. Is this height sufficient for safe and stable flight? Are there known limitations or risks when flying in such confined indoor spaces?
It depends on how you plan to use it: with pilot in the loop or completely autonomously? Technically there is absolutely no problem with 3m, even flying through shelves. Starling 2 runs VIO algorithm for localisation allows no GPS (indoors) operation, although VIO drifts: This is way more than enough if pilot is on the loop, correcting for the drift.
For fully autonomous flight, VIO drift must be compensated; typically using visual references such as AprilTags (already implemented in ModalAI stack) or a SLAM approach (which the algo doing VIO, OpenVINS, is capable of but AFAIK it is not used). Performance depends strongly on lighting, scene features, etc, so real-world testing is essential.
Is the Starling 2 Max suitable for indoor stock control or inventory management applications?
I cannot guarantee but, technically, if you implement the missing algorithms yourself I see no issue.
Are APIs, SDKs, or technical documentation available to support custom integration?
See docs here, how to build native custom apps and ROS apps
Flight controller and onboard software. Which software or firmware comes pre-installed on the flight controller?
FC is ModalAI's branch of PX4.Ardupilot seems also possible. Main controller is ModalAI's flavor of Ubuntu incl. their SDK.
What built-in safety features are available (collision avoidance, emergency stop, geofencing, etc.)?
You have online mapping and path planning, ideal for fully autonomous operation. The main idea is you use the ToF camera depth + VIO to create a map of your surroundings on the fly. Then use this map for path planning. PLEASE note the disclaimer in the docs: This is the one of the most tricky algos to get reliable, so you would have to test intensively and, potentially make SW and HW changes, to guarantee reliable operation.
Is a “return to home” or similar failsafe function available for indoor operation?
Land mode should be possible to configure as failsafe
Any insights, experiences, or recommendations would be very welcome.
Please note this is a development drone intended for companies and research groups, not a finished product.
I hope it helps