Develop AI/ML algorithms for advanced engagement coordination & weapon pairing
NSWCDD is seeking game-savvy students to develop AI/ML algorithms for the automated scheduling & coordination of simulated directed energy, hypervelocity projectiles & other advanced weapon systems.
For the purposes of this prize challenge, “Participant” refers to college or university teams that apply to the prize challenge and “Winner” refers to college or university teams selected to receive a cash prize under this prize challenge.
Naval Surface Warfare Center Dahlgren Division (NSWCDD) is seeking creative game-savvy software innovators from colleges and universities to develop artificial intelligence (AI) and machine learning (ML) algorithms for the automated scheduling and coordination of simulated directed energy, hypervelocity projectiles, and other advanced weapon systems. The training and demonstration of these algorithms will be executed with the Joint Cognitive Operational Research Environment (JCORE), the Navy’s in-house video game. JCORE provides a medium-fidelity simulation of fleet-level exercises and bridges the gap between tabletop wargaming and high-fidelity modeling and simulation to provide a repeatable, faster-than-real-time solution.
The goal of this challenge is to bring participants to solve a problem in a way that NSWCDD has not yet explored. Each participant will be comprised of no more than five student members. The participants will use artificial intelligence and machine learning to design and train algorithms that will trigger their ship to respond to various adversarial actions, demonstrating the advantages of AI/ML integration into our systems and warfighting capabilities. Prior experience with JCORE is not required. Experience with AI and ML algorithm development, Java and Python is necessary.
NSWCDD has been involved in the development of AI/ML algorithms to support enhanced engagement coordination capabilities to optimize Naval warfighting. The Navy continues to investigate AI/ML capabilities to enable rapid and automated weapon pairing. The AI/ML capabilities will enable identification of the best engagement actions and weapons pairings to defeat various notional threats. This will provide NSWCDD with an enhanced solution set to support the automation of tactical decisions such as threat evaluation, combat control, and weapon systems assignment.
This innovation challenge will enable the validation of new and advanced AI/ML algorithms through modeling, simulation and wargaming. It will expand the body of knowledge in AI/ML algorithm design and development, engagement coordination, and hardkill and softkill weapons integration, and enable collaboration across the university ecosystem to support advanced AI/ML research.
The goal of this prize challenge is to develop AI/ML algorithms to expand our solution set(s) for advanced engagement coordination and weapon pairing through a real time innovation challenge event. For the purpose of this challenge, an AI/ML algorithm is defined as an algorithm that automates engagement decisions that can set or adapt its logic paths based on the performance of previous runs. This will advance the state-of-the-art for future automated naval engagements, and increase the speed at which operational decisions are made.
- NSWCDD is seeking to recognize college and university technology student leaders in the areas of software engineering, computer science, artificial intelligence and machine learning, who demonstrate creativity, out-of-the-box thinking, and innovation in their technology endeavors and projects.
- NSWCDD is seeking solution(s) to address the need for automated engagement of weapon to threat pairing in order to most effectively protect naval assets and defeat incoming threats. The impacts of the solutions will enable much more rapid decision making to support ship defense.
This prize challenge will be conducted in two phases. Phase I is an evaluation of technical white papers submitted by participants. Phase II will be an in-person demonstration for selected participants utilizing the Navy’s JCORE modeling and simulation wargame to validate the developed AI/ML algorithms.
Prize Challenge Phase I: White Paper
The objective of Phase I is for participants to demonstrate, via a white paper, the team’s knowledge of advanced AI/ML skills, their ability to develop and apply advanced AI/ML algorithms, and their strategy and plans to approach the Phase II in-person challenge.
Participant white papers must be submitted within the U.S. General Services Administration’s CHALLENGE.GOV website for this prize challenge topic. Entries must be uploaded as either a .PDF, .doc, or .docx document and shall not exceed five pages in length, single sided, using font Times New Roman, 12 or larger. The submission should not include any Personally Identifiable Information (PII) beyond contact information for the Participant and team members (i.e., team member name(s), email addresses and telephone number(s)). The submission must conform to the Terms and Conditions specified in this prize challenge and must address the Phase I criteria below:
Each Participant’s Phase I submission must demonstrate the team’s experience in the development of AI/ML algorithms, and in Java and/or Python programming. The team for the Participant must demonstrate it has the requisite knowledge and skills to develop AI/ML algorithms, and test and train those algorithms in rapid fashion, to meet the objectives of the prize challenge. Submissions must address, at a minimum, the following items:
- Describe the specific AI/ML experience of the team in algorithm design development and application;
- Describe the benefits of the team’s participation in this challenge and how each member plans to use this experience in their future academic and/or career endeavors;
- Describe the team’s approach toward designing, developing, and training AI/ML algorithms for the purpose of enabling automated engagements or actions when encountering various obstacles. This approach should demonstrate the team’s comprehension of the problem(s) to be solved;
- The entry must articulate why the team wants to participate in this challenge, and why they believe they will develop an AI/ML algorithmic solution that will best meet the objectives of this prize challenge. Examples of previous developments and projects are strongly encouraged.
The submission for each Participant must be signed by each member for the Participant’s team and an authorizing person for the college and university. NSWCDD will use a panel of judges to select up to 25 Phase I submissions to participate in the Phase II in-person development and demonstration of AI/ML algorithm development and validation via JCORE modeling and simulation wargaming.
Prize Challenge Phase II: AI/ML algorithm development and demonstration
The objective of this phase is to assess the capabilities and accuracy of the solutions that are developed at the in-person challenge. Selected teams will be invited to demonstrate and apply their AI/ML skills in a live innovation challenge at the University of Mary Washington Dahlgren Campus located in Dahlgren, Virginia.
In this phase, university teams will use artificial intelligence and machine learning to design and train algorithms that will trigger their ship to respond to various adversarial actions, demonstrating the advantages of AI/ML integration into our systems and warfighting capabilities. NSWCDD will provide each Phase II participant with a sample dataset and game Applications Programming Interface (API) before the event to support their understanding of the Challenge and aid in preparation. On the date of the event, access to all software and development environments will be provided. Laptops will also be provided.
Each team will develop an algorithm using AI/ML to defend their friendly forces (1-n friendly ships) against enemy forces of varying sizes and capabilities. During the event, the teams will be provided increasingly difficult supervised training sets to develop their algorithm. These training sets consist of multiple game scenarios that provide the “world state” and current score to the algorithm at each time step and accepts orders from the algorithm to launch effectors or take other actions with their friendly force. After providing any orders, the algorithm will then notify the game that its turn is complete and the game will run until the next time step. Throughout the scenario the teams will need to adapt to changing circumstances as enemy forces launch their own effectors in an attempt to destroy the team’s controlled forces.
The “world state” messages will list the ships they control, their locations, their inventory, list of effectors with their capabilities, and a predicted raid size that may or may not be accurate. The effectors’ capabilities are subject to infrequent change depending on the scenario. At each time step, the team is able to launch effectors, schedule engagements for the future, and cancel any scheduled engagements. The algorithm will be able to use the effectors it has available to schedule engagements against enemy effectors and enemy launchers to launch immediately or in the future.
The game uses Google Protocol Buffers to specify message formats and ZeroMq to connect the algorithm to the game on ports defined by the game. NSWCDD will provide code that can be used within the teams’ algorithms to handle the communications with the game in Java or Python. Alternatively, teams can develop their own interface code in any language supported by Google Protocol Buffers 3.
At the conclusion of the in-person challenge, each participant shall provide NSWCDD with the following artifacts, which will be developed as part of the of the Phase II demonstration.
- Source code of the team’s algorithm with libraries used
- Build instructions for the algorithm to run on the host platform
- Working executable for the host platform
- A brief describing the technical approach used, key challenges conquered, and overview of the results on the test data
All of the above submissions can be provided via removable media (preferably CD/DVD) or can be uploaded to a dedicated site to be provided during the event
Up to three participants will be selected as winners under Phase II.
NSWCDD estimates the following timeline for the completion of the prize challenge.
White Paper Submission:
Opens Monday, July 15th, 2022
Closes Monday, October 31, 2022 by 4:00 PM ET
NSWCDD Challenge Talk:
A virtual panel will be held to answer questions related to the Challenge.gov posting and the Prize Challenge. A firm date for this panel has not yet been set, however it is anticipated that this panel will be held in early September 2022. Interested parties are encouraged to monitor this notice for the official announcement of the virtual panel.
Interested participants are encouraged to submit questions that do not contain any information specific to the team’s approach/experience. If an interested Participant believes a question is necessary that addresses a subject that is specific to the team, the inquiry must be marked accordingly by the submitter. The cutoff for questions sent via electronic mail is October 14, 2022. The challenge will be periodically updated to provide answers to all non-specific team questions. Answers to those questions that are team specific will be responded to directly.
Evaluation of Phase I submissions is tentatively scheduled to occur between November 1 – December 15, 2022. NSWCDD anticipates notifying the selected Phase II Participant teams NLT January 6, 2023.
Thursday, March 2 – Saturday, March 4, 2023
The event will occur at The University of Mary Washington Dahlgren Campus, 4224 University Drive, King George, Virginia 22485.
Winners are expected to be announced on March 5, 2023.
CANCELLATION OF PRIZE CHALLENGE
NSWCDD reserves the right to cancel, suspend, and/or modify the challenge, or any part of it, if any fraud, technical failures, or any other factor beyond NSWCDD’s reasonable control impairs the integrity or proper functioning of the challenge, as determined by NSWCDD in its sole discretion. NSWCDD is not responsible for, nor is it required to accept, incomplete, late, misdirected, damaged, unlawful, or illicit submissions.