Enhancing Cooperative Control with Hierarchical Intelligence and Learning

Period of Performance: 04/20/2012 - 01/22/2013


Phase 1 SBIR

Recipient Firm

Utopiacompression, Corp.
11150 W. Olympic Blvd. Array
Los Angeles, CA 90064
Principal Investigator


ABSTRACT: Intelligent cooperative controller capable of continuous learning from offline and online experience is of sustained interest for a team of for unmanned aircraft to execute complex missions in a dynamic environment. Current cooperative control technologies are lacking of learning capability. Without the ability to learn, a cooperative system may not be able to react to unanticipated scenarios or adapt to dynamically changing environments in a correct and intelligent fashion, and therefore may lead to unsatisfactory system performance or even mission failure. In collaboration with Brigham Young University, UtopiaCompression Corporation (UC) proposes a hierarchical learning framework that enables multi-layer learning. A modular learning process is proposed, which takes as sensor observations, input prior information from offline training, mission objectives from higher level and learned information from lower level, to facilitate intelligent decision making. To demonstrate the feasibility of our proposed framework, we propose cooperative control and learning algorithms that learns to gauge the intent of a target using multiple observations and to guide a tracking mission based on the observed behavior of the target in a dynamic environment. The feasibility analysis and the demonstration will illustrate performance improvement and potential benefits of equipping cooperative control technologies with intelligent learning mechanisms. BENEFIT: In support of effective operations of unmanned aerial vehicles (UAVs) in increasingly complex and uncertain missions, the proposed technology will enable UAVs to learn and adapt to the uncertain environment and changes in adversary behavior. The solution will significantly increase the UAV-to-human ratio for successful operation reducing the overall deployment cost. The proposed learning framework will provide a team of UAVs with a set of tools to react effectively to changing environment, mission objectives and sensor characteristics. The new capabilities will allow the UAVs to complete the missions while flying safely, thus reducing costs due to possible accidents and mission failure. Within the commercial domain, the key technology areas and related applications that can potentially benefit from the proposed technology include surveillance around a critical or secure infrastructure, tracking of unknown targets and classifying their behaviors for border security, search of targets for rescue or surveillance, resupplying UAVs for aircraft carriers, Micro Air Vehicles and flying swarms for reconnaissance and remote monitoring, and civilian search and rescue. All of these applications require intelligent cooperation and decision making between UAVs, which will be enabled through our proposed hierarchical learning framework. UC has identified numerous product opportunities within the US Military modernization effort centering on implementing C4ISR (Command, Control, Computers, Communication, Intelligence, Surveillance, and Reconnaissance) technologies. ISR spending in the next decade is estimated at $15 billion, and depends largely on the stable and efficient control of unmanned vehicles. The UC team is particularly optimistic about the value of integrating UCs technology into the emerging smaller tactical UAVs under development and early deployment such as the RQ-7 and RQ-11 programs. Other potential application programs include the Predator UAV programs, as well as multiple FAA compliance efforts currently underway.