Home | Members | Research | Facilities | Publications | Software | News | Contact


 

   

Intelligent Software Agents

Conflict detection and resolution is a critical capability for the realization of UAV flight operations, particularly in a battle airspace with manned aircraft, and in the Next Generation National Airspace (NGATS), in which all guidance and conflict resolution is performed by onboard systems instead of by ground based control.  A particularly demanding situation within this environment occurs when multiple simultaneous traffic and weather conflicts arise, so full realization of this capability will require advanced systems to assist operators in managing information and decision-making.  While operators will retain ultimate command and control, on-board automation tools will greatly assist operators in various flight tasks.  For example, data link communication devices will collect large amounts of real-time traffic, weather, and terrain information from various sources; a Vehicle Management System (VMS) will monitor vehicle health and compute real-time optimal flight trajectories which comply with mission objectives and a robust, fault tolerant flight control system will fly the UAV.  An advanced decision support aid can integrate these tools and assist operators in processing information, making decisions, and avoiding mistakes.

Although an operator of a UAV may simultaneously encounter more than one type of conflict or hazard in flight, much of the research on decision support aids address only one specific type.  For situations involving multiple flight hazards, such as co-existing thunderstorms and multiple potential traffic collisions, combining several different methods or extending only one method to handle other types of conflict is usually formidable.  This is because each method is usually tailored to address only one specific type of the conflict.  Our approach to address the problem of satisfying multiple objectives of different types is unique, and solves this multiple conflicts resolution problem using a multi-objective fuzzy decision-making method.  It is called the Smart Cockpit Decision Support Aid, and it can operate in either a fully autonomous mode, or a semi-autonomous mode.  It is currently tailored to the enroute flight phase in the NGATS free flight environment for General Aviation aircraft, but it can be easily extended to human operators/controllers of UAVs.  The system objective is to increase flight safety and reduce human related mistakes during the enroute flight phase, by aiding operator decision-making.  The heart of the Smart Operator Decision Support Aid is an intelligent agent based hierarchical system, composed of lower-level intelligent weather and traffic agents which recommend weather and traffic conflict free flight paths respectively, and a higher-level intelligent executive agent that recommends conflict free flight paths in situations of co-existing weather and traffic conflicts.  The executive agent’s capability to resolve multiple conflicts of multiple types is realized by applying the multi-objective fuzzy decision making method.

An important way in which the Smart Operator Decision Support Aid using this approach differs from others reported in the literature, is in the location of the decision making function.  Conflict detectors and solvers are often coupled into a single agent, where an agent is defined here as a module (object) that internalizes all the functionalities necessary to perform some well-defined tasks.  Agents are usually tailored to handle one specific type of conflict, since different conflict types have different characteristics. 

Our system uses multiple agents in an hierarchical system, and is designed specifically for applications in which the responsibility for decision making resides entirely with the UAV remote operator, as opposed to a system like DAG-TM in which the responsibility for decision making is distributed between a centralized ground entity and a remote operator based entity.  To avoid the problem of conflicting solutions between agents, a higher-level agent called the executive agent is created to serve as a mediator.  It is important to note that the executive agent does not process raw mission and flight information, and it does not perform conflict detection and resolution of a specific type of conflict.  On the contrary, based on the post-processed information provided by lower-level agents, the executive agent has the capability and the authority to provide the best resolution for all conflicts.  In our system, the conflict detection and resolution (CD&R) module is a hierarchical multi-agent system (Figure 1), which is composed of multiple sensors feeding data and information to lower-level agents (such as a weather agent and traffic agent), and a higher-level agent (executive agent), which arbitrates and then feeds an  interface manager, which displays de-confliction options to the operator. 
    
It is important to note that the purpose of a decision support aid is to assist operators, not to replace operator reasoning.  Therefore, operators will continue to be involved in the conflict resolution decision-making process.  After the proposed de-confliction strategy is determined by the system, it is displayed to the operator for verification before being executed.  There are two courses of action available to the operator.  The first is to accept the proposed de-confliction and have the controller on the UAV track it and fly it, or have the operator fly it himself.  The second course of action is for the operator to reject the proposed de-confliction, resolve it manually, and then fly it either manually or with the controller on the UAV.       

 

Hierarchical Intelligent Smart Operator Decision Support Aid

Hierarchical Intelligent Smart Operator Decision Support Aid

To validate the Smart Operator Decision Support Aid, we have developed the system and integrated it with an operator interface and flight management system.  Preliminary evaluation has been conducted using real-time flight simulation.  The findings which follow are based upon operator comments, and are preliminary since they are not based on a formal human factors study.  The overall response from all test and evaluation operators was positive, in that situational awareness was perceived to have been improved, and de-confliction workload was perceived to have decreased.  The Smart Operator Decision Support Aid showed operational flexibility beyond the intent of the designers, as some test operators used it as designed, while others used only parts of it during certain portions of the mission, according to their needs and experience level.  The low time operators were glad to have this conflict resolution capability available during what they perceived to be intense situations of combined weather and traffic conflicts (based on their lower experience level).  They allowed the Smart Operator Decision Support Aid to issue flight paths and let the controller on the UAV fly when conflicts were not perceived to be severe, then switched to using the proposed flight paths as references only, manually flying the UAV when conflicts were perceived to be severe.  The high time operators tended to fly under manual control the entire mission, regardless of conflict severity, and only used the Smart Operator Decision Support Aid as an operator associate.  They were glad to have the advisor capability to assist with decision making, but felt that experienced operators such as themselves could probably have worked through the conflicts without it, but with an admittedly higher workload.  They also reported that at no time was a proposed de-confliction obviously wrong or inconsistent with what they themselves generated.  These preliminary findings demonstrate that the proposed approach and Smart Operator Decision Support Aid is a promising candidate for the design of intelligent operator aids and decision-aiding tools for control of UAVs.   

PI: John Valasek, R. Bhattacharya


Related Publications:

•  Rong, Jie, Bokadia, Sangeeta, Shandy, Surya U., and Valasek, John, “Hierarchical Agent Based System for General Aviation CD&R Under Free Flight,” AIAA-2002-4553, Proceedings of the AIAA Guidance, Navigation and Control Conference, Monterey, CA, 5-8 August 2002.
•  Shandy, Surya U., and Valasek, John, “Intelligent Agent for Aircraft Collision Avoidance,” AIAA-2001-4055, Proceedings of the AIAA Guidance, Navigation and Control Conference, Montreal, Canada, 6-9 August 2001.
•  Rong, Jie, Spaeth, Theresa, and Valasek, John, "Small Aircraft Pilot Assistant: Onboard Decision Support System For SATS Aircraft,"AIAA-2005-7382, Proceedings of the AIAA 5th Aviation, Technology, Integration, and Operations Conference (ATIO), Arlington, VA, 26-28 September 2005. 
•  Rong, Jie, and Valasek, John, “Onboard Pilot Decision Aid for High Volume Operation (HVO) in Self-Controlled Airspace (SCA),” Proceedings of the 23rd Digital Avionics and Systems Engineering Conference, Salt Lake City, UT, 24-18 October 2004. 
•  Wollkind, Steve, Valasek, John, and Ioerger, Thomas R., “Automated Conflict Resolution for Air Traffic Management Using Cooperative Multiagent Negotiation,” AIAA-2004-4992, Proceedings of the AIAA Guidance, Navigation, and Control Conference, Providence, RI, 16-19 August 2004.
•  Rong, Jie, Geng, Shijian, Valasek, John, and Ioerger, Thomas, “Air Traffic Conflict Negotiation and Resolution Using An Onboard Multi-Agent System,” DASC-345, 21st Digital Avionics Systems Conference on Air Traffic Management Systems, Irvine, CA, 27-31 October 2002.

 

 



 


© Aerospace Engineering, Texas A&M University