BAA-FA8650-17-S-6001-Call2-SOO.pdf

Statement of Objectives

Autonomy Capability Team 3 Processing Exploitation and Dissemination

(ACT3 PED)

5 April 2018

Background:

Commissioned by the Air Force Research Laboratory (AFRL) Commander in late 2017, ACT3 is an autonomy ‘special operations’ team leveraging an innovative ‘start-up’ development model to advance the science / technology / development of Autonomy. Under its development model, ACT3 is extremely aggressive in the maturation of technology – the team is actively maturing technologies that provide a platform for all AF autonomy (including Autonomy in Motion and Autonomy at Rest). It is defining the shortest path to successful implementation of solutions using AFRL’s internal expertise and collaborations with the best academic and commercial artificial intelligence (AI) researchers in the world.

Objectives:

AFRL will be conducting research and development in AI resulting in transition to operational systems with missions related to processing, exploitation, and dissemination (PED). The ACT3 government experts will work hand-in-hand with industry and academia to develop innovative and operationally relevant capabilities. The scope of the effort will span advanced technology development and will utilize modern information technology (IT) systems capabilities to accelerate development. Industry support can span any or all of the following thrust areas.

Thrust 1 - Algorithm Development: The ACT3 team will develop innovative solutions to PED mission areas, requiring contractor research in the following areas:

• Innovative autonomous systems capable of operations with and without direct human interaction. These autonomous systems will operate on sensor data to provide analytics that human analysts can utilize to create intelligence products. The analyst may also direct the focus on the autonomous functions (e.g., target initiation).

• Autonomous teaming methods that include machine-to-machine, and man-to-machine implementations. These interfaces need to allow human analyst’s to ingest machine generated data and direct future machine processing. Real-time feedback methods are paramount to allow flexibility in these processes. Efforts will include both development of new approaches and analysis of these capabilities in operationally relevant environments.

• Functional capabilities that span the areas of o Target detection, track, ID: multiple AI based methods should be developed or evaluated to automatically generate detection, track and ID of ground targets in various sensor sources. Potential sources include EO and RF sensors from on-site and remote platforms.

o Object correlation and fusion: These functions will provide the ability to ingest multiple sources of the detection, track, and ID capability to generate a cohesive view of the environment. Methods will also enable reduction of clutter/false alarms.

o Entity generation and higher level reasoning: Based on the products created by detection, track, ID, correlation and fusion, high-level reasoning will be developed to generated activities associated with these products (e.g., dismount running), but should also extend to items such as course of action determination.

o Multi-agent processing / management: The multi-agent captures the relationships of Agents (man and machine) and the environment. Here, Agents exist in an organization and connect (communicate) to other Agents. Agents perceive information about the environment and acting upon what they perceive. The environment is dynamic and events will influence and/or change the current state of the environment.

Thrust 2 - Human Machine Teaming (HMT): The ACT3 team will develop and assess autonomous capabilities aimed at PED missions with a focus on HMT and mission effectiveness. Capabilities include, but are not limited to:

• Human-machine interface development and analysis. The focus is to develop novel approaches that utilize the various products under Thrust 1 to help the human analyst generate products.

This involves effective user interfaces, but more importantly, new methods that allow rapid integration of the automated products. These methods need to account for a variety of mission needs such that analysts could focus the product generation based on evolving missions versus a static set of existing missions.

• Algorithm assessment to include failure mitigation analysis. While the automated algorithms will be designed to operate flawlessly, they will encounter failure modes that need to be mitigated through innovative human interaction. New concepts need to be developed and implemented that identify component failure and how the human can interact with component failure.

Assessment of both automated capabilities and failure mitigation methods will be essential to developing effective system level capabilities.

• Operational utility to include metrics on product generation and user workload. Assessments will be conducted at the component level, but must also be conducted at the system level. This includes metrics on overall system utility, improvement in product generation timeliness and accuracy, and reduction in user workload.

Operation Security (OPSEC):

The contractor shall provide OPSEC protection for all sensitive/critical information as defined by AFI 10-701 (Operations Security), the 711 HPW OPSEC Plan, and critical information list. The contractor shall participate in the 711 HPW sustained OPSEC awareness training or include OPSEC training as part of their on-going security program.