Advanced Materials and Manufacturing, ASA(ALT), Phase I

AI/ML for Visual Processing of Energetic Defects

Release Date: 05/12/2022
Solicitation: 22.4
Open Date: 05/26/2022
Topic Number: A224-013
Application Due Date: 06/28/2022
Duration: Up to 6 months
Close Date: 06/28/2022
Amount Up To: 250K

Objective
The purpose of this topic is to utilize Artificial Intelligence and Machine Learning (AI/ML) in conjunction with an in-line process control systems in order to identify defects in energetic fills of munitions including, but not limited to: cracks, voids, gaps, foreign material and chemical agent leakage. Proposal should leverage existing vision and process control system technology and energetic defect characterization studies to detect, define and decide in real time to eliminate defective parts leaving a production floor. The objective is to develop a high accuracy vision system capable of being scaled to images ranging from primers to small caliber to artillery sized energetic billets, with adaptable power to penetrate various packaging materials.

Description
In current times, energetic filled parts are inspected for defects during manufacturing processes utilizing x-ray equipment. Critical defects are inspected 100%, especially for items such as Excalibur, in support of the LRPF CFT using a pass/fail criteria. Each load plant has at least basic x ray capability, and Armaments Center has lab scale X ray and CT capability but neither meet the required need to find, identify and mark for culling any critical defects.

AI/ML paired with a visual processing system will allow for efficient, correct identification of defects in energetic fills and assembly. AI/ML which builds upon energetic defect modeling will allow production plants to properly identify critical defects which cannot be sent to the field. Rejected parts will be culled from manufactured lots, reducing potential for incidents in the field due to undetected defects

Overall, a Visual system paired with a trained AI/ML model can be inserted as an in-line step in all energetic manufacturing without adding significant delay to manufacturing. Proposal should integrate a scalable visual processing control system, capable of correctly and repeatedly identifying defects in energetic fills, ranging in size from primers up to a 155mm energetic billet, with an AI/ML algorithm which identifies defect type and severity for culling from production lots. Defects presently include, but are not limited to: cracks, voids, gaps, foreign material and chemical agent leakage.

Phase I
Provide feasibility study to ensure all safety and material handling requirements have been addressed for utilizing a vision system in conjunction with energetic materials.

Phase II
Develop lab scale visual processing system capable of consistent and repeatable energetic defect detection at correct position to adequately capture defect (up to 50 mm energetic fills); Develop database of defects correlated to imaging data for several energetic items; Create and train lab scale models to identify defects for several end items.

Phase III
Scale up lab scale system to pilot (up to 105mm) and then production scale (up to 155mm) for in-line defect detection in manufacturing scale processes while maintaining high resolution at necessary speed and scale

While the explosive nature of this topic makes it niche, the visual inspection of primers allows for applications in mining, food packaging, and microelectronics.

For the actual submission dates and to submit your full proposal package, visit the DSIP Portal.

AI/ML for Visual Processing of Energetic Defects

References:

  1. Engel, W., Herrmann, M., 2001. Lattice Imperfections of Energetic Materials Measured by X Ray Diffraction. Defense Technical Information Center Technical Report from Fraunhofer Institut fur Chemische Technologie
  2. Baker, E., Sharp, M., 2018. Gun Launch and Setback Actuators, 2018 Insensitive Munitions & Energetic Materials Technology Symposium Portland, OR; Munitions Safety Information Analysis Center (NATO), Brussels, Belgium
  3. Trujillo, D., Guziewski, M, Coleman, S., 2019. Machine Learning for Predicting Properties of Silicon Carbide Grain Boundaries; Defense Technical Information Center Technical Report from Army Research Laboratory

Objective
The purpose of this topic is to utilize Artificial Intelligence and Machine Learning (AI/ML) in conjunction with an in-line process control systems in order to identify defects in energetic fills of munitions including, but not limited to: cracks, voids, gaps, foreign material and chemical agent leakage. Proposal should leverage existing vision and process control system technology and energetic defect characterization studies to detect, define and decide in real time to eliminate defective parts leaving a production floor. The objective is to develop a high accuracy vision system capable of being scaled to images ranging from primers to small caliber to artillery sized energetic billets, with adaptable power to penetrate various packaging materials.

Description
In current times, energetic filled parts are inspected for defects during manufacturing processes utilizing x-ray equipment. Critical defects are inspected 100%, especially for items such as Excalibur, in support of the LRPF CFT using a pass/fail criteria. Each load plant has at least basic x ray capability, and Armaments Center has lab scale X ray and CT capability but neither meet the required need to find, identify and mark for culling any critical defects.

AI/ML paired with a visual processing system will allow for efficient, correct identification of defects in energetic fills and assembly. AI/ML which builds upon energetic defect modeling will allow production plants to properly identify critical defects which cannot be sent to the field. Rejected parts will be culled from manufactured lots, reducing potential for incidents in the field due to undetected defects

Overall, a Visual system paired with a trained AI/ML model can be inserted as an in-line step in all energetic manufacturing without adding significant delay to manufacturing. Proposal should integrate a scalable visual processing control system, capable of correctly and repeatedly identifying defects in energetic fills, ranging in size from primers up to a 155mm energetic billet, with an AI/ML algorithm which identifies defect type and severity for culling from production lots. Defects presently include, but are not limited to: cracks, voids, gaps, foreign material and chemical agent leakage.

Phase I
Provide feasibility study to ensure all safety and material handling requirements have been addressed for utilizing a vision system in conjunction with energetic materials.

Phase II
Develop lab scale visual processing system capable of consistent and repeatable energetic defect detection at correct position to adequately capture defect (up to 50 mm energetic fills); Develop database of defects correlated to imaging data for several energetic items; Create and train lab scale models to identify defects for several end items.

Phase III
Scale up lab scale system to pilot (up to 105mm) and then production scale (up to 155mm) for in-line defect detection in manufacturing scale processes while maintaining high resolution at necessary speed and scale

While the explosive nature of this topic makes it niche, the visual inspection of primers allows for applications in mining, food packaging, and microelectronics.

For the actual submission dates and to submit your full proposal package, visit the DSIP Portal.

References:

  1. Engel, W., Herrmann, M., 2001. Lattice Imperfections of Energetic Materials Measured by X Ray Diffraction. Defense Technical Information Center Technical Report from Fraunhofer Institut fur Chemische Technologie
  2. Baker, E., Sharp, M., 2018. Gun Launch and Setback Actuators, 2018 Insensitive Munitions & Energetic Materials Technology Symposium Portland, OR; Munitions Safety Information Analysis Center (NATO), Brussels, Belgium
  3. Trujillo, D., Guziewski, M, Coleman, S., 2019. Machine Learning for Predicting Properties of Silicon Carbide Grain Boundaries; Defense Technical Information Center Technical Report from Army Research Laboratory

Soldier shooting in the field

AI/ML for Visual Processing of Energetic Defects

Scroll to Top