On-Controller Drone AI: Running Inspection Software on DJI Enterprise Controllers
The controller is where flight decisions already happen. Adding mission and detection tools there can simplify field work—if compute, interface and data paths are carefully designed.
Quick answer
On-controller drone AI places mission software, live overlays or local analysis on the pilot’s controller or connected device. It can reduce extra field equipment, keep tools close to flight control and preserve local records. SAI-HUB RC runs in professional DJI controller workflows and can also connect to Brain-Box for broader on-premises processing and storage.
Why the controller is a useful edge location
The controller already receives video and telemetry and is operated by the person who can change or stop the mission. A controller-based application can present route design, target selection, detection overlays and flight records in one field interface. That shortens the cognitive and physical path between observation and response.
It also avoids carrying a separate analytics laptop for some missions. The tradeoff is finite compute, screen space, battery and platform compatibility. Safety-critical flight information must remain clear even when AI produces many detections.
- Create waypoints, orbits, grids and patrol routes in the field.
- Switch between manual flight and supported automated missions.
- Choose detection profiles for the intended RGB or thermal feed.
- View live overlays and stream selected video to a control room.
- Retain mission and detection records for post-flight review.
Controller inference, on-board inference and Brain-Box compared
These locations are complementary. On-board AI processes closest to the camera but follows the aircraft platform’s model pipeline. Controller inference receives the transmitted feed and keeps the pilot interface local. Brain-Box can add more compute, storage and integration with other sensors or systems.
Choose task by task. A Matrice 4 on-board detector might surface a fast cue; the controller presents and acknowledges it; Brain-Box retains the mission and runs secondary workflows. Or a non-Matrice-4 feed may be analyzed locally by controller or Brain-Box instead.
| Location | Best for | Key check |
|---|---|---|
| Aircraft | Immediate compatible detection | Model and device-deployment constraints |
| Controller | Portable pilot-facing mission and AI workflow | Supported hardware, resources and interface clarity |
| Brain-Box | On-premises storage, integration and broader processing | Stream path, sizing and site infrastructure |
What SAI-HUB RC adds to professional DJI operations
SAI-HUB RC is Spectro AI’s remote-controller configuration for supported DJI enterprise and selected controller workflows. It combines manual and autonomous mission control, waypoint and pattern planning, live RTMP streaming, AI model selection and local flight or detection logs.
Spectro AI lists support across Matrice, Matrice 4, Matrice 30, Mavic 3 Enterprise and selected Mini series with compatible controller arrangements. Because aircraft, controller and firmware combinations evolve, buyers should validate the exact set they intend to operate.
How to design the pilot interface around attention
An alert that covers telemetry or produces constant noise can make a mission less safe. Limit active classes to the objective, set thresholds using representative data and distinguish an informational overlay from an alert that requires action. Provide one clear route to inspect the original scene.
Autonomous patrols should make manual takeover obvious. If a target appears, SAI-HUB RC can support a workflow in which the route pauses and the operator decides whether to investigate or continue. Train this transition until it is routine.
- Show essential flight state before analytics detail.
- Use class, confidence and time to explain each detection.
- Prevent duplicate alerts from the same tracked object.
- Make pause, manual takeover, continue and return actions unambiguous.
- Record the operator’s disposition for later model review.
Field acceptance tests for controller software
Test in sunlight, gloves, cold or heat and the network conditions of the real site. Measure battery impact, sustained inference, screen responsiveness and storage. Interrupt the internet and video link separately; they are different failures. Verify that core flight behavior stays within the supported DJI design.
Finally, export a complete mission. Confirm that media, detection summaries, coordinates or GIS outputs and operator notes can be retrieved by the people who need them. A controller application succeeds when it reduces field complexity without trapping evidence on one screen.
- Outdoor readability and touch interaction.
- Latency, frame rate and interface responsiveness.
- Manual/autonomous transitions and return procedures.
- Offline behavior, local capacity and later synchronization.
- Log completeness, export and role-based access.
Frequently asked questions
Can AI object detection run on a DJI controller?
Yes, compatible applications can analyze the video feed or display detections at the controller. Performance depends on controller hardware, application design, model and input resolution.
Which controllers support SAI-HUB RC?
Spectro AI lists DJI RC N, RC Pro and RC Plus families in supported arrangements. Confirm the exact controller, connected device, aircraft and firmware before deployment.
Does SAI-HUB RC store flight data locally?
Spectro AI describes local storage for video and detection records, including summaries and supported geospatial outputs. Confirm retention and export requirements for the selected configuration.
Can controller video be viewed in a control room?
SAI-HUB RC includes live RTMP streaming capability. Remote viewing depends on the configured network and should be secured, bandwidth-tested and included in outage procedures.