IN Brief:
- Palladyne AI plans to integrate SwarmOS collaborative autonomy with NORDA Dynamics' Underdog terminal-guidance module.
- The architecture separates distributed detection, tracking, and task allocation from short-range pursuit and engagement by individual aircraft.
- The companies plan integration and joint demonstrations, but no customer, contract value, platform selection, or demonstration date has been disclosed.
Palladyne AI and NORDA Dynamics plan to integrate two autonomy systems covering different phases of an uncrewed-aircraft mission, linking Palladyne’s SwarmOS collaborative-control software with NORDA’s Underdog terminal-guidance module. The companies intend to install the combined architecture on one or more aircraft and conduct joint demonstrations, subject to any necessary government approvals.
SwarmOS is intended to handle distributed perception and coordination across several uncrewed platforms, including detection, classification, task allocation, and persistent tracking of multiple targets. Underdog addresses the later part of the sequence, providing short-range pursuit and terminal guidance after a target has been assigned. The proposed interface would pass target-state information from the swarm-management layer directly into the guidance software controlling an individual aircraft.
That division reflects the different processing problems involved. Coordinating several aircraft requires a broader view of tracks, available platforms, task priorities, and the behaviour of other members of the group, while terminal guidance has to react quickly to changing geometry close to the assigned target. Combining the two requires the systems to agree on target identity, coordinates, timing, and mission state as control shifts from collaborative allocation towards individual guidance.
Both companies place particular emphasis on degraded communications. Palladyne describes SwarmOS as an edge-based architecture that distributes decision-making between participating systems rather than requiring every action to pass through a single central controller. NORDA says Underdog is designed to provide navigation, recognition, and terminal guidance when GNSS or communications are unavailable, addressing environments where remote piloting or conventional satellite navigation may be disrupted.
The collaboration also retains human oversight around engagement decisions. According to the joint announcement, operators remain human-on-the-loop while the software handles more of the detection, tracking, assignment, pursuit, and guidance chain. That distinction leaves rules of engagement, abort authority, identification thresholds, and acceptable autonomy behaviour as part of the system architecture rather than treating automation as an unrestricted replacement for command decisions.
NORDA’s own operational evidence requires some caution. The Ukrainian-Estonian company says its technology is used by more than 200 units in Ukraine’s defence forces and reports 93% mission effectiveness for the Underdog 0.8 update based on analysis of more than 1,000 mission reports. Those figures are supplied by NORDA and have not been independently validated in the material released with the partnership, so they establish the company’s claimed operating experience rather than a separately verified performance benchmark.
Palladyne brings a different body of test experience. Earlier in August, SwarmOS was used during the US Army’s Project Convergence Capstone 6, where the company says it coordinated aircraft from different manufacturers while integrating with Anduril’s Lattice and the Army’s developing command-and-control architecture. That work concentrated on heterogeneous drone coordination and command-system integration; the NORDA collaboration extends the architecture towards terminal guidance rather than repeating the same demonstration.
A separate 19 August agreement with Orkid Tech also covers the integration of Palladyne Pilot and SwarmOS with a fixed-wing VTOL aircraft. Taken together, the agreements indicate that Palladyne is trying to keep its autonomy software independent of a single airframe manufacturer. Platform independence can widen the potential market, but every new aircraft still requires interfaces to flight controls, sensors, payloads, power systems, communications equipment, and safety functions to be engineered and tested.
The same problem applies to the NORDA integration. SwarmOS and Underdog may operate at different stages of a mission, but the boundary between them has to function when communications are intermittent and target information is changing. Latency, stale tracks, coordinate errors, conflicting sensor assessments, and lost links all require defined behaviour; an architecture intended for electronic-warfare conditions cannot assume the clean data paths available during a laboratory demonstration.
Computing and power requirements also move further onto the aircraft as autonomy becomes more distributed. Sensor fusion, target recognition, path planning, coordination, and guidance all consume processing capacity while competing with payload and communications equipment for electrical power, cooling, and weight. Software assurance becomes more difficult at the same time because several interacting algorithms can produce system behaviour that is harder to test than a fixed sequence of remotely commanded actions.
The 26 August announcement does not yet contain evidence that the combined architecture has flown. No customer, contract value, production quantity, selected aircraft, or joint demonstration date has been named, and the companies explicitly make future integration subject to government approval where required. The next meaningful milestone is therefore straightforward: a flight demonstration showing SwarmOS handing target information to Underdog reliably while several aircraft operate under degraded communications.
Until that test occurs, the development remains an integration programme rather than a fielded combined system. Its technical interest lies in the interface being attempted — connecting distributed multi-aircraft perception and tasking directly to autonomous terminal guidance. If the hand-off works under the communications conditions the companies are designing around, that interface will carry more weight than either company’s broader claims about autonomy on its own.
