What to expect at AUSA 2026
See edge AI across the systems operators already use
The Association of the United States Army’s Annual Meeting returns October 12–14 to the Walter E. Washington Convention Center in Washington, D.C. This year, you’ll find our technology running live inside six partner environments across three venues.
At AUSA, we’re showing what it looks like when edge AI becomes part of the systems operators already use: different platforms, sensors, compute environments, and mission applications. Across the show, you’ll see Latent AI technology running on a drone platform, ruggedized edge compute, and a low-light imaging system.
The hardware and integration change, but the underlying AI capability stays portable.
Visit our AUSA 2026 event page
The perception and targeting layer for any system
Latent AI ATR is the perception and targeting layer that runs underneath your existing platform. It combines foundation-model tracking with persistent target memory, so it can cue any target from a click, AI detection, plain-language description, or target photo, then maintain the track through motion, lighting changes, and occlusion. It can also share target memory across platforms to support handoff and maintain custody through a mission.
ATR runs onboard the edge hardware already part of the platform, including Nvidia, NXP, and Qualcomm, rather than relying on the cloud. No retraining. No customer training dataset required. The tracking layer can integrate with existing detection, VLM, manual cueing, targeting, ISR, or autonomy workflows without requiring operators to redesign the platform around a new AI stack.
At AUSA, you can see that capability integrated into different platforms, sensors, and mission environments:
Wind River X (Westin, Floor 2, Blossom Room)
The integration brings together Latent AI, Systel, and Wind River Linux to demonstrate Latent AI C-UAS solutions running on Wind River Linux on the Kitestrike II.
- What you’ll see: Latent AI ATR and C-UAS solutions running on an edge compute device with WRX Linux as the embedded operating system. ATR will also demonstrate automatic gimbal control driven by the tracking module. Why it matters: AI running at the edge needs to work across different hardware and operating environments. Tracking can feed downstream mission applications, from drone autonomy to gimbaled weapon systems, while C-UAS detection and threat analysis can happen locally without reachback.
One Stop Systems — Hall D, Booth #7454
At One Stop Systems, two Latent AI capabilities come together on OSS’s Gen 5 3U SDS-LC short-depth server, demonstrating how AI workloads can run on ruggedized edge compute built for demanding environments.
Latent Analyze will run on the OSS Gen 5 3U SDS-LC, demonstrating how video analytics can help physical security teams investigate recorded footage without sending data back to a centralized environment. Visitors can use natural-language queries to search and investigate video archives.
You can also see Latent AI ATR running from a laptop and camera, showing how the tracker cues and maintains a target without requiring a full drone platform.
- What you’ll see: Natural-language queries used to search and investigate recorded video, alongside a live demonstration of ATR tracking a cued target. Why it matters: Security teams collect video from multiple feeds, but reviewing that footage manually can take hours. Latent Analyze helps teams find and understand relevant events faster, while ATR demonstrates how persistent tracking can run on a lightweight edge setup.
Leonardo DRS — Hall D, Booth #6341
See Latent AI’s C-UAS perception capabilities hosted on the Leonardo DRS THOR Edge Compute solution, demonstrating how AI-powered detection and tracking can operate directly within a mission-ready edge architecture.
- What you’ll see: Latent AI C-UAS detection and tracking running on the Leonardo DRS THOR Edge Compute solution, with persistent tracking of small UAS targets. Why it matters: C-UAS systems need reliable perception close to the mission, where bandwidth, connectivity, and compute resources can be constrained. This demonstration shows how Latent AI’s edge perception layer can detect and track targets without relying on centralized processing.
Army GVSP
GVSP demonstrations will run from 9-11am on Monday the 12th, 12-2pm on Tuesday the 13th, and 3-5pm on Wednesday the 14th in the drone flight area.
- What you’ll see: Latent AI ATR running on an NVIDIA Jetson AGX ORIN, 64Gb, detecting and tracking two different drones flying in the GVSP demo area.
- Why it matters: Real-time perception has to run on the compute available on the platform. This demonstration shows ATR tracking multiple targets directly on an embedded NVIDIA Jetson platform during live flight operations.
Qinetiq (SRI table) — Hall C, Booth #3417
At the SRI table in QinetiQ’s partner pavilion, see SRI’s DomiNite® digital night-vision and QinetiQ US anchor the Army’s shift from analog to digital soldier architecture. Latent AI’s ATR evolves this setup from passive viewing to precise, human-authorized engagement through a unified tactical interface—supporting everything from human-on-the-loop “click-to-lock” to fully autonomous operations.
- What you’ll see: SRI DomiNite low-light video stream processed live on a tactical tablet by the Latent AI ATR engine. Tap any target in the dark camera feed, and the system instantly paints a persistent digital mask that locks onto and tracks the moving object through pitch-black surroundings.
- Why it matters: We demonstrate a key capability for autonomous operation in ultra-low-light missions: real-time, persistent tracking on a low-SWaP edge device with zero network connectivity. This solves the core tactical ISR challenge: holding a flawless lock on moving threats at the edge when light and bandwidth are scarce.
Sigma Defense — Marriott Marquis, Adams Morgan Room M3
At Sigma Defense, see Latent AI’s edge solutions running across multiple edge computing platforms, including PacStar, Kägwerks Dock Ultra, and CardShark, with sensor feeds integrated through Sigma’s NEOS edge hardware.
- What you’ll see: Latent AI’s edge solutions running on Sigma’s edge hardware and pulling live sensor feeds through NEOS, demonstrating how AI capabilities can be deployed across different compute platforms and integrated into an existing sensor architecture.
- Why it matters: Edge AI has to work within the hardware, sensors, and network environments already used in the field. This demonstration shows how a common AI layer can operate across different edge devices while connecting to sensor data through an existing integration architecture.
Six integrations. One portable AI layer.
These demonstrations span different hardware, sensors, and mission environments, but they share a common idea: AI should work with the systems already in the field. That means running at the edge, working within real system constraints, and integrating with the platforms and sensors already in use. At AUSA, you can see Latent AI running across different compute platforms, sensor feeds, and mission systems. The hardware and mission change, but the AI layer can move with them.
- ATR, multi-target tracking, and automatic gimbal control through Wind River X
- AI-powered video intelligence and ATR on ruggedized edge compute through One Stop Systems
- Multi-target drone tracking on NVIDIA Jetson through Army GVSP
- C-UAS detection and persistent tracking on the Leonardo DRS THOR Tactical Edge Compute solution
- Low-light target tracking with DomiNite through SRI
- Multi-platform edge AI and sensor integration through Sigma Defense
See Latent AI at AUSA 2026
Find us across six partner environments at AUSA 2026 and see how Latent AI brings edge perception to different hardware, sensors, and mission systems.
Recent Blogs