Before friend or foe, a tactical AI must be able to say: “Unknown.”
Baoqi Smart has completed the first engineering slice of a new Project SHEEPDOG research capability for identity under uncertainty.
In mixed, communication-degraded environments, forcing every observed contact into an immediate friendly-or-hostile label can create false confidence. SHEEPDOG is being designed to preserve ambiguity when the available evidence is insufficient, request additional verification, and keep the reasoning available for replay and review.
The larger product direction is straightforward: coordinate sensing and mission resources according to what the system knows, what it does not know, and what needs to be verified next—without sacrificing coverage, reserve or human oversight.
Why this matters
Visually similar platforms, incomplete communications and mixed civilian environments make identity a decision problem, not just an image-recognition problem. Premature certainty is itself a mission risk.
This work extends SHEEPDOG beyond multi-vehicle route planning toward explainable tactical decision support across heterogeneous autonomous systems.
Research boundary: this is a deterministic synthetic research capability under development. It is not operational IFF, target identification, autonomous interception or autonomous engagement. Consequential real-world actions remain subject to explicit human authorization and platform-level safety controls.
