Docstoc

ITEE seminar Sep 05

Document Sample
ITEE seminar Sep 05 Powered By Docstoc
					       Dynamic Weather Avoidance
    Trajectories in a Traffic Constrained
              Enroute Airspace
Minh Ha Nguyen, Sameer Alam, Jangjung Tang , Hussein Abbass

                The Artificial Life and Adaptive Robotic Laboratory,
        University of New South Wales at Australian Defence Force Academy
                        http://www.itee.adfa.edu.au/~alar

  The 6th EUROCONTROL Innovative Research Workshop & Exhibition




Minh Ha Nguyen   Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 1/20
                                                                                    Outline
                     Introduction
                     Algorithm Design
                           Constraints and Objectives
                           Problem Definition
                           4D Grid Design
                           Safety Inherent Pre-processing
                           Conflict Free Weather Avoidance
                     Ant Colony Optimization
                     Experiments and Results
                     Conclusion
Minh Ha Nguyen   Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA      Paper #18, 2/20
                 Introduction

      Weather is the single largest contributor to delay in the
      air traffic control system and is a major factor in
      aircraft safety incidents and accidents (E.g.: 70% of all
      delays in U.S. National Airspace)

      Weather disturbances can also severely damage the
      airframe of the aircraft. They can potentially damage
      the navigational and electronic equipments leading to
      pilot’s loss of control.


Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 3/20
                 Algorithm Design
The key objectives in developing a weather avoidance
  algorithm for an enroute traffic constrained airspace are:
 •Ability to search multiple routes in the presence of multiple
 airspace constraints - including fast moving convective weather cells
 & neighbouring air traffic.
 •Ability to handle flight optimization objectives: e.g. minimizing
 changes in heading, altitude and deviation from planned trajectory.
 •Take into consideration aircraft performance parameters.
 •Ability to generate weather free trajectories such that there is no
 loss of separation with neighbouring traffic.
 •Provide the set of optimal solutions to choose from.

Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 4/20
                 Multiple Constraints
      Performance constraints:
            Maximum operating altitude
            Climb/descent rate
            Maximum permissible turn, govern the degrees of freedom
             available for weather cells avoidance manoeuvres.


      Airspace Hazard Constraints: these hazards are
      present in the airspace when a particular region is inaccessible
      for manoeuvring, either reserved for special use (SUA) or that
      a particular manoeuvre may lead into conflict situation with
      other neighbouring aircrafts in the airspace.

Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 5/20
                 Multiple Objectives

      Avoid weather cells
      Minimize deviation from flight plans
      Minimize heading changes
      Minimize climb/descent manoeuvres




Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 6/20
                  Problem Definition
Given:
  a 4-D grid of dimensions I (latitude) x J (longitude) x K (altitude) x
T (time)
  an entry point x (start manoeuvre point)
  an exit point y (end manoeuvre point)
  locations of weather cells and their severity in this grid
 find the routes between x and y on the given objectives satisfying
the problem constraints.




 Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 7/20
                 4D Grid Design
     The algorithm detects weather cells at                                     Is there a weather
                                                                                                           No
     60nm (weather radar range) & generates a                                    cell within 60nm?
     3D volume of 100nm x 100nm x 3000ft                                                     Yes
     around the cells.                                                         Generate a 3D block
                                                                             of 10nm x 10nm x 3000ft
                                                                                around weather cells
     It is then discretized into a grid of 10 x 10
     x 3 (300 nodes)                                                                 Discretize into
                                                                                     10x10x3 grid


     The 3-D grid is extended to 4-D by                                          Take snapshot of
                                                                                 block at every t
     including time as the 4th dimension.
           The simplest way to encode the time                                      Write to 4D grid
            dimension is to take a snapshot - aircraft
            positions and weather cells’ positions and
            intensities - of the discretized grid at t time              Find entry point         Find exit point
            interval
                                                                                      Call ACO to
                                                                                     generate routes
Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA              Paper #18, 8/20
                 Safety Inherent Pre-processing

    Eliminate the state space recursively starting from an
    entry point in the 4-D grid and doing forward
    recursion on different layers.

  At each layer, eliminate states which violate the
  problem constraints before moving to the following
  layer.
 Thus, we can guarantee that the resultant state space
  contains feasible transitions only.


Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 9/20
                 Conflict Free Weather Avoidance

   Weather free routes are
computed and converted into
grid-cell plans.

   If the ETAs of the
computed aircraft and of the
traffic aircrafts, at a cell
common in their routes, are
within 5 minutes  a
collision is signalled and the
route is discarded

Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 10/20
                  Ant Colony Optimization
                                                                    ACO is based on the
                                                                    foraging mechanism
                                                                    employed by real ants
                                                                    attempting to find a
                                                                    shortest path to a food
                                                                    source.

Ants use indirect
communication via the
environment by employing
pheromone trails
 Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 11/20
                 Experiments
 Middle-east airspace is used for weather modelling and
   flight simulation.
 Two complex weather patterns are used as test problem:




          Clustered pattern                              Distributed pattern
Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 12/20
                 Result 1: Clustered Weather




    t0                    t1                       t2                      t3               t4


Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 13/20
                 Result 2: Distributed Weather




    t0                   t1                      t2                       t3              t4


Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 14/20
                 Performance Constraints




  Weather routes generated by ACO                Rate of Climb/Descent is between
                                                    15m/s




Rate of Acceleration is between 0.7m/s2 Rate of Turning is between
  Minh Ha Nguyen                                                  3deg/s
                 Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA Paper #18, 15/20
            Conflict Free Weather Avoidance

                                                  FLT031 detects weather cells and generates
                                                         Weather Avoidance Routes


                                                FLT031 updates its intent information on airspace


                                                 FLT032 detects weather and intent information
                                                           of other traffic aircrafts


                                                   FLT032 generates Weather Avoidance and
                                                            Conflict Free Routes


                                                FLT032 updates its intent information on airspace



                                                  FLT032 & FLT031 fly safely through airspace



Minh Ha Nguyen   Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA       Paper #18, 16/20
            Result 3: Conflict Free Weather Avoidance



                  t0                                       t1                                t2




      t3                               t4                                       t5



Minh Ha Nguyen   Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA    Paper #18, 17/20
            Result 3: Conflict Free Weather Avoidance




                 Separation in terms of distance and altitude
            (Distance between two aircrafts  21.44 nautical miles)

Minh Ha Nguyen   Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 18/20
                 Conclusion

      A safety inherent design utilizing intent
      information can provide a good framework
      for dynamic conflict free weather avoidance
      Pre-processing the state space before search
      helps in reducing the search space
      ACO successfully generates optimal flyable
      trajectories for different weather patterns in a
      constrained airspace.

Minh Ha Nguyen    Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 19/20
Minh Ha Nguyen   Artificial Life & Adaptive Robotics Laboratory (ALAR), UNSW@ADFA   Paper #18, 20/20

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:3
posted:6/28/2011
language:English
pages:20