Over the past two years, artificial intelligence has rapidly evolved from a frontier technology into practical, deployable capabilities that enhance productivity and operational efficiency.
Last year, DJI Enterprise opened up onboard intelligent computing across multiple drone platforms, including the M4T and M400, enabling developers to deploy AI algorithms directly on drones and bring intelligence closer to real-world operations.
DJI Enterprise officially announces the Drone Onboard AI Challenge 2026, an initiative designed to explore, validate, and scale the value of onboard AI in drone-based applications.
Challenge Overview
The challenge is open to developers and users worldwide, inviting participants to explore and unlock the potential of drone onboard AI and accelerate its application across real-world operational scenarios. By enabling AI algorithms to run directly on drones, the challenge aims to drive practical value creation in fields such as power utilities, public safety, inspection, agriculture, natural resources, and emergency response.
Algorithms on Aircraft
DJI opens its onboard computing capabilities to developers. With just a few simple steps of model training and deployment preparation, AI algorithms can be deployed directly onto the latest DJI Enterprise drone platforms, enabling intelligent detection and analysis aligned with real-world operational scenarios.
Manifold 3
Manifold 3 is a computing expansion module designed for DJI Enterprise drones. It provides high-performance edge computing resources and native PSDK interface support, enabling seamless integration and stable communication with drone platforms. By accessing onboard media and sensor data, and when combined with flight control and route planning capabilities, Manifold 3 helps developers build automated and intelligent onboard AI applications.
Algorithms on the Cloud
Based on DJI FlightHub 2's high-definition, low-latency media streaming channel, Algorithms on the Cloud enable rapid integration between live drone streaming and self-built cloud-based intelligent inference services. Users can remotely and in real time view intelligent algorithm detection results and trigger automatic alerts, making drone operations smarter and more efficient.
Competition Schedule (February 2026 – June 2026)
February: Opening of the registration and consultation channel; March: Opening of proposal submission and release of detailed competition guidelines; May: Announcement of shortlisted proposals and opening of trial access and voting channels; June: Proposal evaluation and announcement of award-winning solutions.
Awards and Benefits
Shortlisting Stage: Finalist Award (multiple winners) — Finalist solutions will be showcased on the official DJI Developer competition website. Final Stage: Best Onboard AI Model Award — 5 winners (Osmo Pocket series gift set, Manifold 3 kit, Trophy); Industry Application Excellence Award — 10 winners (Manifold 3 kit, Trophy). Winning solutions will be featured on DJI Enterprise official channels and included in the official 'Onboard AI Solutions Catalog'.