Swarm & Search AI Challenge Rules & Terms

This is a competition to find the best Artificial Intelligence(AI) or Machine Learning (ML) algorithms to accomplish a published scenario task with the greatest efficiency. The Air Force Research Laboratory (AFRL), in parallel with the United Kingdom Defense Science and Technology Laboratory (DSTL) has designed this competition to become more complex as it progresses through the various phases.  They are looking for resilient solutions that would represent real world conditions. 

Any proprietary software that is used in the competition will remain the sole property of the competing teams. Decisions about the validity and merit of participation and submissions in these competitions is at the sole discretion of WBI.  WBI reserves the right to refuse participation or recognition whenever it would not meet the intent of the competition.

  • Pre-Competition Phase 15 Jan – 26 Feb 2019.
  • Online Competition Phase 12 Feb – 12 Mar 2019.
  • Onsite Showdown Phase 29 Mar – 31 Mar 2019 in Dayton, Ohio.
  • Highest scoring solution featured at AUVSI Conference 30 April – 2 May 2019 in Chicago, Ill.
  • Highest scoring solution featured at U.K. Embassy in Washington DC, June 2019

PRE-COMPETITION PHASE

Practice phase scenarios will be simplified and allow teams to get familiar with the software.  The development team estimates 1-2 days for acclimation. 

  • The team will upload run file(s) from their practice runs and their score files will be verified.
  • The scoring algorithm will be measures of task performance, with an accounting for resources committed or consumed to accomplish the task.  
  • The verified scores will be sent back to the team lead.  The teams can submit as many run files as they desire to improve their best score.
  • A scoring algorithm will be provided in the AMASE software to provide unverified score outcomes during algorithm development.
  • For questions about, or help with AMASE, please contact joe.althaus@wbi-innovates.com

ONLINE COMPETITION PHASE

Around 12 February 2019 the final competition scenario will be presented to the registered teams. This scenario will be used to determine the final ten teams to be invited to the March showdown.

As the competition progresses the basic scenario will be similar, but the following parameters may be introduced or varied:

  1. Winds may become variable within the scenario
  2. The fire zone and smoke zone dynamics may vary
  3. Number of fire zones and their location could be changed
  4. No fly zones or high-risk zones may be introduced
  5. The task may be expanded to require identification of High Value Targets at risk
  6. The number and types of UAV available may change.
  7. Sensors may be slewable and their performance may vary
  8. Communication between UAVs and command algorithm may be limited or inhibited.
  9. Other changes disclosed ahead of the competition
  • Teams will develop their strategy and starting conditions for their autonomous runs to accomplish the tasks in the competition scenario. 
  • The Team Lead will upload their output AMASE run file and score file to the competition website where it will be verified.
  • Once verified, the team score will automatically post to a leader board, unless they fall below fifteenth place.
  • Scores below 15th place will be sent to the team lead but not posted on the leader board. 
  • Teams can submit as many run files as they desire to optimize their scores.

ON-SITE SHOWDOWN PHASE

On 15 March the online competition will close.

  • The top ten selected teams from the U.S. will be invited to attend the on-site showdown, in Dayton, Ohio.
  • The U.K. competition will create their own, separate on-site event.
  • The Showdown will commence with a joint opening session of the two participating countries.
  • All teams will be given the same scenarios with additional complexities and the same time limit for each “heat.” 
  • Teams may be invited to make a short presentation on their approach and the overall results will be shown to the attendees.  
  • The highest scoring solutions will be decided from the outcomes from all “heats.”  

This competition will be run using the Air Force Research Laboratory developed AMASE software. The software (Open AMASE, available UAVs, Sensors, and use information), a simple learning scenario, a video tutorial, and a demo program will be available for download.  Teams are not allowed to distribute the Open AMASE to anyone outside of their team. An online Help Desk forum will be available for questions on running the competition or issues with the distributed software.

Read and download the AMASE Unilateral License.

US teams must have a US Citizen as a team lead who will be responsible for communication with the rest of their team and the competition coordinators.  Other team members do not need to be US citizens.  Team Leaders must attest to their citizenship to be registered in the closed access site (APAN), and later provide proof of US citizenship to be eligible for prizes.   

Participant Expectations

  1. Selected teams from the online competition are expected to participate in the showdown competition, in Dayton Ohio on the 29th -31 March 2019 with a small group (no more than 5).
  2. The team with the highest scoring solution agrees to be highlighted at the AVUSI conference held in Chicago from Apr 30-May 2 2019, as well as, send a representative to the UK embassy for special recognition (expenses reimbursed).
  3. Showdown teams agree to be filmed and included in a highlight video that may be published on an AFRL website.
  4. Showdown teams agree to discuss their approach and strategies for this competition with the AFRL autonomy experts.
  • Participating teams will have the chance to network with AFRL autonomy experts and discuss a range of opportunities anticipated in 2019 and beyond.
  • The top U.S. teams at the showdown will be awarded cash prizes, totaling more than $50K, based on performance in each heat, overall performance, and exceptional recognition. They will also be provided with up to $5K in travel reimbursement (limit 5 team members) to attend in-person.
  • High-scoring solutions will be recognized by AFRL at the AUVSI’s Xponential trade show in May and a winning representative will be invited to the UK Embassy in Washington DC for special recognition in June. 

Leave a Reply

Your email address will not be published. Required fields are marked *