AIR TRAFFIC CONFLICT NEGOTIATION AND RESOLUTION USING AN ONBOARD MULTI-AGENT SYSTEM, John Valasek, Shijian Geng, and Thomas R. Ioerger 21 st Digital Avionics Systems Conference Irvine, CA October 2002
GENERAL AVIATION FREE FLIGHT Primary goal is to enable free flight for General Aviation (GA) A new way of managing air traffic Departure from the highly structured system to a more flexible system Pilot allowed to change route, speed and altitude, within limits More responsibility for aircraft separation safety rested on Pilots Ultimate decision making authority lies with the air traffic controllers Modern techlogies make Free Flight possible 21 st DASC (2)
CONFLICT DETECTION AND RESOLUTION (CD&R) Critical factor for realizing Free Flight Detects potential conflicts, e.g.,weather, traffic, etc. Recommends pilots with optimal, conflict-free flight paths In this paper, a conceptual CD&R is proposed, using a pair-wise argument-based negotiation approach Application of Multi-agent System (MAS) techniques Collaboration among aircraft for searching multilateral acceptable conflicts resolutions Optimal resolutions with respect to certain cost functions Real-time constraints are considered Previous research on CD&R utilizing negotiation method Wangermann and Stengel s Aircraft/Airspace System Haper et. al., principle negotiation among aircraft 21 st DASC (3)
MULTI-AIRCRAFT AGENT SYSTEM (MAAS) MAAS Concept Weather, Traffic, Terrain Information Negotiation Aircraft Agent ADS-B Device Tracker Conflict Detection Flight Plan Management FMS + CD&R Negotiation Flight Control System Conflict Resolution 21 st DASC (4)
OVERALL SYSTEM DESIGN Definitions: Flight plan, Segment, and Proposal Assumption: Aircraft class: FAR 23 class GA aircraft Advanced on-board devices: ADS-B device, Flight Management System (FMS), communication suites, and data link devices 3 Degree of pilot automy Only automy level 3 are considered Mixed-Fleet Environment ATC represents aircraft lack of negotiation capacities Degree Of Pilot Automy (Harper s Definition) Different on-board devices, while same responsibilities and priorities among aircraft Agent Properties The MAAS is n-deterministic, mixed-event, dynamic All agents have incomplete kwledge, are maximizing, rational, and cooperative 1 2 Standard ATC. Pilots act as instructed by ATC Pilot is free to search for and negotiate potential solutions with other pilots of level 2 or higher and with ATC, and implement the resulting globally approved actions Pilot is free to search for and negotiate potential solutions with other pilots of level 2 or higher, and pose solutions to ATC for approval before implementation 21 st DASC (5)
CD&R MODULE Logic Flow Control for CD&R Module Weather, Traffic, Terrain Information Conflict Resolution Conflict Detection Conflict detected? Any traffic conflict? Deadline expires? Conflict Resolution & Negotiation Solution Approved Turn to ATC Solution created? Solution Implementation Solution provided by ATC Turn to ATC 21 st DASC (6)
CD&R MODULE (cont.) Conflict Detection Weather, terrain conflicts and Special Usage Airspace are represented as restricted areas in 3-D airspace Traffic conflict detection takes ADS-B vectors, determined by protected and alert zones Conflict Resolution A constraint satisfaction problem Resolution space: composed of all possible flight plans Constraints: posed by various conflicts, flight rules, aircraft configuration limits, etc. Cost functions: arrival time, fuel consumption, passenger comfort, etc. Protected Zone Alert Zone Conversion of the resolution space to a discrete, limited domain 3 maneuvers: changing headings, airspeeds, or altitudes Conflict resolution composed of discrete maneuvers with specified starting and ending times 21 st DASC (7)
CD&R MODULE (cont.) Argumentation-base negotiation protocol 1 Send proposal Any new proposal reduces cost? Send Acceptance Send Rejection Any new proposal satifies all constraints Proposal satisfies all constraints? Communication with other aircraft Communication with ATC 1 Initiator starting point Any acceptance Any (counter) proposal received? 2 Responder starting point 2 Logic Flow of the Negotiation Between Two Agents 21 st DASC (8)
CD&R MODULE (cont.) Greeting process Before the real negotiation, initiated by the aircraft first detecting the conflict Confirming conflicts, exchanging on-board devices information, and determining initiator of the negotiation Pair-wise strategy for group conflicts Difficult to determine a multilateral negotiation process in a dynamic MAS system Multi-way conflicts are resolved by a series of pair-wise negotiations ATC plays role in assuring global airspace safety A problem requires more research effort 21 st DASC (9)
PRELIMNARY SYSTEM IMPLEMENTATION A simple multi-agent simulation system is created Individual Aircraft Agent As described in the third slide A state-based negotiation module 0 Waiting 4 0 1 6 5 7 1 Receiving a proposal 2 Receiving an acceptation 3 8 2 3 4 5 6 7 8 Receiving a rejection Ready to send a proposal Ready to send an acceptation Ready to send a rejection Reaching a mutual agreement Unable to find a resolution to the conflict Simulated Squall Line Environment of MAAS Simulated severe weather conditions, terrain variations 21 st DASC (10)
TAMU FLIGHT SIMULATION LAB Fixed-base: Rockwell Commander 700; AV-8A Harrier SGI Onyx Reality II sim engine Networked bank of PC s Center stick; sidestick 155 o projected field of view 30 Hz refresh rate Head Up Display (HUD) 21 st DASC (11)
TAMU FLIGHT SIMULATION LAB (cont.) Head Down Displays (HDD) Reconfigurable CRT; touchscreen LCD Autopilot Glide slope capture Heading Altitude Pitch attitude Flight Management System (FMS) Jeppesen data base Pre-flight planning and enroute updating Moving map display 21 st DASC (12)
NUMERICAL EXAMPLE Head-on CD&R sim B (a) Before Conflict (a) sim A (b) (b) Starting Avoidance (c) Conflict Resolved (d) Back to Original Path (c) (d) 21 st DASC (13)
NUMERICAL EXAMPLE (cont.) Head-On Conflict Negotiation Process sim_b: I suggested a conflict exists between sim_a and me sim_a: sim_b suggested a conflict exists sim_a: sim_b has conflict resolution module sim_a: I do have conflict resolution module, I may propose first sim_a: I suggested a conflict exists between sim_b and me sim_b: sim_a suggested a conflict exists sim_b: sim_a will propose first sim_b: I confirmed the conflict with sim_a, and sim_a will go on proposing a avoidance plan sim_a: sim_b confirmed the conflict, and I will go on proposing a avoidance plan sim_a: I propose a conflict avoidance plan to sim_b sim_b: sim_a proposed an conflict avoidance plan sim_b: I accept sim_a's proposed plan sim_a: sim_b accepted my proposed plan sim_a: I confirmed sim_b's acception of the plan sim_b: sim_a confirmed my acception of the plan 21 st DASC (14)
NUMERICAL EXAMPLE (cont.) If Both Aircraft Without Conflict Resolution Capability sim_b: I suggested a conflict exists between sim_a and me sim_a: sim_b suggested a conflict exists sim_a: sim_b has conflict resolution module sim_a: I do t have conflict resolution module either,turn to ATC sim_a: I confirmed the conflict with sim_b, but I am t able to propose a plan either sim_b: sim_a confirmed the conflict, but it is t able to propose a plan either, turn to ATC 21 st DASC (15)
CONCLUSIONS A multi aircraft agent system is proposed for GA free flight environment CD&R for each agent is solved as a constraint satisfaction problem A pair-wise, argument-based negotiation approach is established searching for multilaterally acceptable conflict resolutions A simple multi-aircraft simulation system demonstrate the proposed inter-agent negotiation method to be a promising candidate for CD&R Only conceptual and preliminary efforts is presented More advanced air traffic scenarios More complicated mixed fleet environment More types of vehicle systems 21 st DASC (16)