ZAĞANOS
KREOSUS · ZAĞANOS ROBOTICS · SUAS 2026
Kreosus is our rapidly deployable storm-response quadrotor, engineered to map damage, detect targets and deliver aid within one mission clock.
A fifteen-second visual introduction to Zağanos Robotics and our first SUAS mission in 2026.
ABOUT US
Zağanos is a multidisciplinary technology team developing projects in unmanned aircraft, underwater robotics, artificial intelligence and competitive robotics.
We are bringing the experience earned across national and international competitions to our first SUAS season. For Tulsa, our aircraft, avionics, autonomy, perception and payload teams are working toward one shared flight system.
KREOSUS · VERIFIED DESIGN
The current competition aircraft combines a 4.922 kg quadrotor, bounded V6X flight authority, ground-assisted RF-DETR perception and two independently gated delivery winches. Its public TDR links every claim to an evidence type.
Explore the aircraft and evidenceOUR UAV JOURNEY
The milestones that brought Zağanos' autonomous aircraft experience to its first SUAS season.
1st in Eurasia · 3rd in the world
Our autonomous drone team represented Zağanos on the international stage.
1st in Türkiye · TUSAŞ Special Award
Zağanos UAV completed the season with the national championship and the TUSAŞ Special Award.
Our first SUAS season
We are carrying this experience into the international SUAS competition on 14–17 September 2026.
Republic of Türkiye Ministry of Youth and Sports
Meet the people bringing Zağanos Robotics' first SUAS mission to life.
Team Coordinator
Team Advisor
Team Member
Mechanics
Team Member
Mechanics
Team Member
Electronics
Team Member
Software
Team Member
Project Management
Team Member
Mechanics
Team Member
Software
Team Member
Mechanics
Team Member
Software
Team Member
Software
Team Member
Electronics
Team Member
Mechanics
Team Member
Electronics
Team Member
Software
Team Member
Project Management
Team Member
MechanicsSUAS 2026 · ENGINEERING RECORD
A requirement-linked record of the decisions, tests, failures and remaining gates that produced the current competition configuration.
The five scored mission categories were converted into verification thresholds before the final hardware configuration was retained.
The Safety Pilot owns visual line of sight, takeover and abort. The GCS Operator monitors mission state, and manual target selection remains disabled while armed.
The 4.922 kg quadrotor was designed for the 18 × 14 × 8 in Personal Item and a tool-free ready time of no more than 180 seconds.
The V6X alone commands propulsion. LTE carries imagery and signed intents that remain subject to V6X and RP2040 local gates.
Mass, deployment, hover efficiency and governing structural loads were closed on the current competition configuration.
The navigation ceiling is 19.0 m per second and the 60 percent command cap retains 2.02 total thrust-to-weight at competition mass.
All ten deployments passed 180 seconds. Print-axis material cards returned a minimum directional factor of safety of about 2 in the governing cases.
The selected propulsion produced 1,230 gf per rotor at 85.0 W hover and 2,489 gf per rotor at the software cap.
The complete perception path was evaluated using actual 150 ft AGL flight data and a full 84-image mission-load transfer.
RF-DETR Nano achieved 96.4 percent precision and 92.0 percent recall on 87 labeled flight-data instances, outperforming the YOLO11n baseline.
The 2028 × 1520 camera mode gives a 73.7 × 55.2 m footprint and 3.64 cm per pixel ground sampling distance at 150 ft AGL.
All 84 images transferred with zero drops while telemetry and RTK corrections remained active. Exposure-to-intent P95 was 1.18 seconds.
Mechanism faults, false-contact signatures and flight delivery accuracy remained visible in the evidence record.
The fourteen-predicate release baseline rejected all 16,383 invalid combinations and produced no unsafe transition in 100,000 randomized event sequences.
Two independent winches completed 197 of 200 cycles. One jam and two wind-slack signatures were detected and contained, with 200 of 200 correct home indications.
All 60 releases above 150 ft AGL landed within 50 ft. Mean radial error was 4.2 m and no uncontrolled release occurred.
Mission phases are linked by documented energy and time gates in the current competition configuration.
All 100,000 gated modeled missions finished within 45 minutes. The tenth lap was authorized in 99.37 percent and modeled P95 remaining energy was 10.7 percent.
Mapping and delivery flights provide representative-altitude inputs for the current mission model.
MAP-001, DET-001, COM-001, REL-SW-001, WINCH-001 and DROP-001 provide traceable evidence across the mission thread.
The current design, decisions, failures and evidence types are public while the team continues final competition preparation.
Flight software, mission gates and release logic remain tied to the test identifiers and evidence reported in the public TDR.
Timed deployment and both delivery channels will be repeated after the final travel cycle and reassembly.
Final preparation focuses on safe full-workflow rehearsals in the competition configuration.
KREOSUS · SUAS 2026 AIRCRAFT
A 4.922 kg quadrotor designed around rapid deployment, tightly bounded flight authority, ground-assisted perception and independently gated aid delivery.
KREOSUS SYSTEMS
Explore the competition aircraft in 360 degrees and select a system to inspect its verified components and measured data.
DEP-001 · measured
PROP-001 · bench measured
MAP-001 · DET-001 · COM-001
Locally bounded architecture
WINCH-001 · DROP-001 · measured
ENGINEERING DECISIONS
Selected alternatives were retained only when they improved mission score, contained failure or reduced verification burden.
The nominal mission targets all ten endurance laps without sacrificing mapping and delivery when time or energy margin is low.
The selected structure reaches the Personal Item transport target without accepting permanent hinge play.
Ground compute removes airborne GPU mass and preserves the two-operator mission structure.
A single jam cannot remove both delivery opportunities and an LTE command cannot independently authorise release.
TEST AND VALIDATION LOG
MEASURED
147.5 s mean · 157.8 s P95 · 160 s worst · no tools
BENCH-MEASURED
1,230 gf at 85.0 W hover · 2,489 gf at cap · 65 °C maximum
MODELED
100 percent within 45 min · lap ten 99.37 percent · P95 40.37 min and 535.5 Wh
FLIGHT AND SYSTEM
150 ft AGL orthomosaic · 84 of 84 transferred · all three planted targets detected
FLIGHT-DATA MEASURED
96.4 percent precision · 92.0 percent recall · 118 ms
SYSTEM-MEASURED
Zero drops · queue depth 3 · exposure-to-intent P95 1.18 s
BENCH-MEASURED
197 of 200 complete · 200 of 200 correct home · detected failures contained
FLIGHT-MEASURED
60 of 60 within 50 ft · 4.2 m mean · 8.4 m maximum · zero uncontrolled releases
ANALYTICAL PASS
4.25 and 27.48 MPa · 0.028 and 0.842 mm · directional factor of safety about 2
SOFTWARE-VERIFIED
All 16,383 invalid states rejected · zero unsafe randomised transitions
AIRCRAFT DATA REGISTER
Values below come from the public 29 July 2026 Kreosus Technical Design Report with their documented test or analysis context.
TEAM BLOG
Verified design updates, flight evidence and the remaining path from the Kreosus baseline to Tulsa.
Our 2026 Technical Design Report now documents the complete aircraft baseline and its evidence boundaries.
The report connects mission requirements to trade studies, measured tests, flight results and failures that changed the design.
A 150 ft AGL sortie produced the stitched MAP-001 orthomosaic and found all three planted targets.
RF-DETR Nano was evaluated separately on 87 labeled instances from actual flight data so the three-target map demonstration was not treated as the full quantitative dataset.
Both independent channels completed thirty flight deliveries in winds from 2 to 8 m per second.
All 60 payloads landed within 50 ft with a 4.2 m mean radial error. The finite perfect sample is reported with its uncertainty rather than as proof of perfect underlying reliability.
KREOSUS GALLERY
Current CAD views and selected figures from the public 2026 Technical Design Report.
We will keep this page aligned with verified tests, design decisions and final competition preparation.
