Artificial Intelligence/Machine Learning,
Machine Learning (ML) for Breach Routing
Release Date: 11/16/2021
Open Date: 11/30/2021
Application Due Date: 01/04/2022
Close Date: 01/04/2022
Topic Number: A214-048
Duration: Up to 6 months
Amount Up To: $250K
This program aims to improve the Explosive Breacher (EB) by closing capability gaps in current breaching techniques left by quickly evolving adversarial technology and aging legacy equipment. Artificial Intelligence (AI) and Machine Learning (ML) algorithms will be trained to identify obstacle locations within a minefield. From this information, the algorithms can create efficient firing plans to feed into EB in order to quickly neutralize threats and determine the best path through the field. This allows vehicles and personnel to safely navigate through the breach as efficiently as possible.
The purpose of this topic is Develop an ML algorithm for threat detection that can be used to create a firing plan for routing through complex enemy minefields; improve effectiveness of EB; modernize breaching for the 21st Century. Currently, effective methods for route planning and breaching enemy minefields are lacking, creating a need for improvements. Legacy breaching equipment, such as MICLIC, is no longer effective against quickly evolving adversarial technology and can be time consuming to deploy. AI/ML will be used to identify explosive obstacles, route through minefields, and create efficient firing plans for breaching technology. AI/ML can analyze data and create solutions more quickly and accurately than personnel with the same information. If successful, capability gaps in current breaching techniques will be closed, increases efficiency and speed of breaching operations, shorter time for soldiers in breach, resulting in reduced likelihood of casualties, and success will be measured by Soldier feedback of the capability, timing of breaching operations, and % of area cleared.
Phase I consists of the development of a prototype AI detection method that can be used to analyze complex minefield information by training an ML algorithm based on obstacle attributes, expected minefield setup patterns related to known obstacles, and performance of the breaching technology to be used among other factors. Demonstrating the feasibility of the algorithm for accurate threat detection is required.
Phase II consists of the maturation of the ML algorithm from Phase I based on performance in laboratory experiments and improved knowledge. It will also require integration with detection system and demonstration of the matured algorithm in relevant environments.
Success of Phase II would allow for integration into the Explosive Breacher program of record and further commercialization into Phase III.
For more information, and to submit your full proposal package, visit the DSIP Portal.
- Douros, C., Looney, E. “Standoff Activated Volcano Obstacle Emplacements Analysis” U.S. Army Research, Development and Engineering Command, October 2018.
- AI/Machine Learning for Minefield Clearance, Vic Grout, https://vicgrout.net/2020/02/25/ai-machine-learning-for-minefield-clearance/