CCC outlines a 2035 roadmap for civilian drone computing
The Computing Community Consortium has outlined a research roadmap for non-military drone computing, with experts identifying 12 technical challenges on the path to 2035. The priorities include closing the simulation-to-reality gap, building shared testbeds and developing safer real-time autonomy.
- CCC’s roadmap addresses 12 technical challenges for drone computing.
- The vision covers non-military drone applications through 2035.
- Researchers called for digital twins built from shared real-world data.
- The USDA Forest Service uses drones for wildfire, forestry and inspections.
- Speakers said humans will remain in the loop as drone AI develops.
A vision for drones in 2035
The roadmap was discussed during CCC’s November 3, 2026, Community Chat, titled Computing on the Fly: Navigating a Vision for the Future of Drone Computing. The event followed a December 2025 workshop attended by 47 experts from academia, industry and government.
The workshop focused on non-military uses ranging from urban and indoor delivery to medical supply transport and wildfire detection. In one wildfire scenario, networks of drones would monitor hundreds of thousands of acres, identify emerging risks and help direct firefighters.
Christopher Stewart, a lead author of the workshop report, highlighted four of the 12 challenges: making software tested on simulated fleets behave the same way in the field, creating reproducible integration testbeds, developing infrastructure to identify aircraft in shared airspace, and building autonomy frameworks that can adapt as a mission expands from 5 drones to 50.
From high-fidelity simulations to digital twins
Sabine Brunswicker, founder of Purdue’s AIrTonomy program, argued that research should move beyond high-fidelity simulation toward digital twins. These models represent specific physical systems, receive real-time data and can run “what-if” analyses during flight, according to the CCC blog.
Examples discussed during the chat included an engine digital twin that detects anomalies using sparse sensor data, language models constrained to the controlled grammar of air traffic control, and a browser-based simulation environment used in a recent competition. Brunswicker invited laboratories, testbeds and test sites to share data and models so researchers can build these environments collaboratively.
The discussion also connected digital twins with access to research. Shared environments could help more institutions participate in drone experiments, as those projects are becoming harder for mid-size universities to run.
Safety, data and the human role
Dirk Giles, National UAS Branch Chief at the USDA Forest Service, described the agency’s standards-based approach to training, operations, airworthiness and pilot qualification. Its drones support wildfire response, timber work, hydrology, reforestation and infrastructure inspection. During fires, they provide thermal views for locating spot fires and are being used for aerial ignition, a task the agency has conducted for more than 50 years with low-flying helicopters.
The Forest Service also processes and stores drone imagery under federal records requirements. That data feeds AI tools for bridge-inspection change detection and nursery-seedling counts, with 100% accuracy for the youngest plants, according to the blog.
Security, supply chains, safe coordination with piloted aircraft and responsible spending were among the issues raised in the question-and-answer session. On AI-driven drones, Giles said a human will remain in the loop for the rest of his career. The discussion therefore frames near-term progress around tools that support pilots and operators, rather than removing them from decisions.
Computing Community Consortium Blog: Simulation to Reality: Takeaways from CCC’s Drone Computing Community Chat
