AIR AutoPilot is the first physical AI powered by a foundation Visual Language Model (VLM) for continuous subject tracking. Built from the ground up for REMI, live events, and broadcast — a force-multiplier for your crew, giving them advantages they can’t find anywhere else.
Physical AI Powered By A Foundation Visual Language Model
Knows what it's tracking — not just motion.
Pan, tilt, and optical zoom with no more image degradation.
Scale coverage without scaling crews.
Manage cameras remotely at ultra-low latency.
Traditional camera autotracking relies on computer vision — rigid, error-prone, and easily fooled by crowds, similar-looking subjects, or fast action. AIR AutoPilot takes a fundamentally different approach.
Using 300 million different data points, AIR AutoPilot has semantic understanding of a scene that makes it more accurate than any other tracking technology. It knows what it's looking at — not just where a shape is moving. That means it can differentiate between players wearing the same uniform, maintain tracking through occlusions, and adapt in real time.
Available as an add-on to the AIR One camera robot, AIR AutoPilot gives you flexibility molded to your situation. For REMI workflows, your team can manage multiple units entirely from distance at ultra-low latency.
No other camera autotracking gives you full wide game-follow mode and individual ISO tracking from the same bank of robots. Your camera operators become camera conductors, focused on framing the shot and telling the story….
The whole-picture wide shot, handled by the robot. AIR AutoPilot follows the field of play with smooth movement at speeds fast enough to keep up with live action, and at the zoom level you determine.
Lock a dedicated robot onto a single subject — athlete, presenter, performer, vehicle — and let AIR AutoPilot follow with the same precision and smoothness as a skilled camera operator.
Not limited to just optical tracking, AIR AutoPilot can also be enhanced by third-party tracking data for even greater accuracy. Get in touch (link to contact page) to discuss your needs.
AIR AutoPilot tracking Golden State Warriors’ Steph Curry through a crowded court and multiple oclussions with zero drops and no latency. In this example, the robot is in San Francisco, the operator is in San Diego, and the AI is in the cloud.
AI tracking using a Visual Language Model with a semantic understanding of the scene allows for camera robot "self driving".
Behind the Scenes, Operator POV -- A single camera operator manages two AI-driven AIR camera robots at the 2026 MLS All-Star Game Halftime Show, featuring Ciara
UI view of AIR AutoPilot conducting ISO tracking during SDFC v AUS
Behind-the-scenes view of 3 AIR camera robots tracking a hockey game on Feb 5, 2026 from a distance of ~120 feet. One camera is following all 10 skaters, one camera is ISO following an individual skater, and one camera is following the puck.
Talk to our team about how AIR AutoPilot fits your REMI workflow, live event setup, or broadcast infrastructure. We'll walk you through a live demo tailored to your production.
Contact Us
