Skip to content

Research Area 3: Autonomous Target Recognition

  1. RW is interested in investigating all aspects of Automatic Target Recognition (ATR) / Autonomous Target Acquisition (ATA) / Aided Target Recognition (AiTR) / Autonomous Target Reacquisition (ATR) technology as it applies to seekers for conventional guided weapons. Interests range from basic signal and image processing foundations through tower and flight test of advanced, real-time ATR/host signal processor implementations. Technical approaches in the areas of pattern recognition, computer vision, deep learning, machine learning, autonomous systems, and cooperative systems as they apply to weapon seekers are of interest.
  2. The following technologies and research areas are of particular interest:

1) AI/ML and/or traditional algorithms for weapon seeker target acquisition or re-acquisition.

2) Investigations and analyses of AI/ML and/or traditional algorithms leading to a better fundamental understanding of their operation and limitations; especially with respect to ATR/ATA/AiTR/ATR applications.

3) Approaches for real-time/on-line training or adaptation of AI/ML and/or traditional algorithms.

4) Approaches for training AI/ML or traditional algorithms with synthetic target data that result in good target recognition performance when using real target data (e.g. synthetic to real domain adaptation).

5) Approaches for cooperative/collaborative ATR using multiple lower cost networked weapon seekers.

6) Approaches for the compact representation of target appearance information.

7) Approaches for automatic/autonomous handoff of target cue information from intelligence, surveillance, and reconnaissance (ISR) or fire control sensors to weapon seekers to improve the ability of the weapon seeker to acquire or re-acquire the target selected by the ISR or fire control system.

8) Methods or tools for the assessment, evaluation, or prediction of ATR performance.

9) Methods or tools for the assessment, evaluation, and analysis of data representations across sensor modalities.

10) Methods or tools for predicting the signature of a target in one sensor domain given its signature in a different sensor domain (e.g., view with synthetic aperture radar [SAR] sensor and predict signature in IR).

11) Approaches to use/incorporate scene context (provided by an ISR or fire control system) for target re-acquisition by a weapon seeker.

12) Approaches to perform image processing, computer vision, or ATR functions directly using compressively sensed image data before (or instead of) image reconstruction.

13) Technologies, research, or approaches that integrate weapon, ISR, and/or fire control subsystems to provide greater overall kill effectiveness, shorter overall kill timelines, lower overall costs, reduced operator burden, and/or greater system autonomy. Topics in this area may be pursued in partnership with other AFRL Technology Directorates (e.g., Sensors Directorate).

14) Software and/or hardware approaches that fully automate the image ground truthing process and provide approximate pixel-level target/background labeling of data sets. The process could be implemented as part of the data collection process or as a post-collection process.

15) Algorithms, or integrated software and hardware approaches that develop or demonstrate improved performance of target detection, classification, or identification algorithms provided by cooperative, collaborative, networked, and/or swarming weapons.

16) Measurements of material properties relevant for use by signature prediction codes in the infrared spectrum (e.g., using DIRSIG) or Ku/Ka frequency bands (e.g., using Xpatch) for more accurate prediction of target signatures in this spectrum / at these frequencies. Additionally, target models (for ingestion by signature prediction codes) that contain model components with accurately typed material properties for more accurate prediction of target signatures.

17) Algorithms or integrated software and hardware approaches that develop or demonstrate alternative navigation capabilities. This may include approaches for radar-aided navigation, celestial-based navigation in a form-factor relevant for munitions, and other non-GNSS (global navigation satellite system)-based navigation approaches.

 

Keywords: Artificial Intelligence/Machine Learning (AI/ML); Target Recognition; Seekers; Non-GNSS based Navigation; Pattern Recognition; Computer Vision; Deep Learning; Autonomous Systems; Cooperative/Collaborative Systems.

 

Contact Us: techtransfer@doolittleinstitute.org