SOFWERX, in collaboration with the USSOCOM PEO-SOF Digital Applications (SDA) Unmanned Systems Autonomy and Interoperability (UxSAI) Program, will host an Assessment Event (AE) to identify technology providers capable of delivering cutting-edge computer vision capabilities for detection and classification for all USSOCOM unmanned systems. This event aims to evaluate whitepaper responses and down-select innovative solutions for a computer vision inference engine and model training solution that can autonomously detect, classify, and adapt to new targets and environments ultimately enhancing the operational effectiveness of UxS in resource-constrained and communication-denied environments to align with the goals and objectives of the UxSAI Program.

Computer vision rarely performs as intended and results in missed detections or improperly classified objects. Challenges exist with obtaining training data, utilizing the models on constrained resources, and deploying models over the air. UxSAI requests computer vision developers to improve our artificial intelligence and machine learning capabilities and develop highly effective computer vision inference engines and pretrained models as part of an Enterprise machine learning operations (MLOPS) pipeline. Providers selected for participation will receive additional information and funding to prepare for the AE upon selection. The UxSAI Program intends to evaluate solutions through their Enterprise MLOPS pipeline as part of this event.
The UxSAI Program seeks a computer vision inference engine and model retraining solution that:

Demonstrates robust object detection and classification capabilities.
Optimizes Size, Weight, Power, and Cost (SWaP-C) for deployment on a variety of UxS platforms.
Enables training and adaptation to new objects and environments.
Offers a modular architecture suitable for seamless integration with existing architectures.
Enhances the autonomous capabilities of unmanned systems operating in challenging environments.
Informs the necessary interfaces, protocols, and data formats for integration, contributing to the development of an Interface Control Document (ICD).
The intent from this event is for UxSAI to work with providers selected from the AE to further develop their model through collaboration with the UxSAI Program

Submit via events.sofwerx.org/uxsai-cvmodeldev NLT 13 October 2025 11:59 PM ET
ITAR Restricted