EURO2013 Creation of Conflict-Free Taxi Routes For Aircraft With Evolutionary Algorithms

www.DLR.
de Chart 1
I. Gerdes, M. Schaper Creation of Conflict-free Taxi Routes for Aircraft with Evolutionary Algorithms EURO|INFORMS MMXIII
Creation of Conflict-free Taxi Routes for

Aircraft with Evolutionary Algorithms
I. Gerdes, M. Schaper
EURO|INFORMS MMXIII, Rome, 30.June - 4. July 2013
www.DLR.de Chart 2
Taxi Routing on Airports: Todays Situation
Aircraft ground movements are guided and controlled by Air Traffic Controllers
(Apron / Ground).
The controllers issue instructions when to start taxi and which route to take.
Controllers carry out conflict detection and resolution.
The mental plan of the controller is nearly nowhere visible to any system.
Taxi speeds are not determined and on pilots decision.
Commands to the pilots are transmitted via radio.
The speeds vary in different areas of the airport (lower on apron, higher on
other taxiways).
www.DLR.de Chart 3
Taxi Routing on Airports: Future Vision

All aircraft have pre-planned, conflict free, time-based taxi routes each including
a speed profile.
Conflict detection and resolution is done automatically.
Route and time conformance is monitored and route/times are adapted for all
moving aircraft.
Controllers are supported by advisories to guide the aircraft as planned.
Clearances are given by data link.
Aircraft are able to follow speed advisories with absolute values (taxi 20 knots
instead of expedite taxi).
Combination of all planning tools like arrival and departure manager (AMAN /
DMAN), stand and gate manager (SGMAN) and surface manager (SMAN).
www.DLR.de Chart 4
TRACC
Taxi Routes for Aircraft: Creation and Controlling
TRACC is a Surface Management System supporting time-based taxi.
Calculates optimized and confict-free taxi routes with speed
recommendations.
Supports the Air Traffic Controller to archieve the proposed time-based
routes.
Adapts planned routes in case of deviations, e.g.
Controller wants other route.
Pilot deviates accidentally.
www.DLR.de Chart 5
TRACC-Approach: Creation of a Conflict-free and Optimized

Taxi Routes
Based on former work about using evolutionary
algorithms for free-routing of aircraft inside the
TMA (ROGENA).
The airport structure is presented as a system of
nodes (waypoints) and links (taxiways).
A waypoint holds information about position of
this point and the standard speed.
Taxi routes are a sequence of waypoints where
each waypoint holds additional information
(planned speed, holding time and time the
waypoint is reached).
Links can be considered as a vector in space:
= + + (+ )
www.DLR.de Chart 6
Workflow of TRACC
www.DLR.de Chart 7
Optimization Concepts
Implementation of two different optimization Algorithms:
TOA (Time Optimization Algorithm):
Usage of a pre-defined standard taxi route and adaption of taxi speeds and
holding times for the creation of a new conflict free route.
The probability for inserting a holding time depends on the difficulties to
create a conflict-free route (Safe-Node).
ROA (Route Optimization Algorithm):
Modification a the complete taxi route referring to the run of the route,
speed profile and holding times.
Holding times are created with a predefined probability from the start.
For both algorithm a special operator flats the speed profiles by removing speed
changes within a defined distance.
www.DLR.de Chart 8
Evolutionary Algorithm of TRACC

Based on modGA of Michalewicz.
Chromosomes are taxi routes with waypoints as genes and additional
parameters.
Trying to prevent super-individuals from overtaking the population.
Population selected for the next generation is divided in 3 different groups:
Group of unchanged routes (including a fixed number of best and different
chromosomes)
Group of chromosomes which will undergo mutation.
Group of chromosomes which will undergo crossover.
Evaluation function depends on the number of conflicts with other aircraft,
number of speed changes, difference to expected time at route destination
(target time) and length of route.
Selection probability depends on the evaluation value of the chromosome.
www.DLR.de Chart 9
Problem Dependent Operators: TOA
Mutation of speeds:
Small probability for each waypoint (gene) for a speed change.
Crossover of speeds:
Selection of a pair of chromosomes.
Selection of two indices as crossover-points.
Exchange of speeds between selected pairs of chromosomes.
www.DLR.de Chart 10 I. Gerdes, M. Schaper Creation of Conflict-free Taxi Routes for Aircraft with Evolutionary Algorithms EURO|INFORMS MMXIII
Problem Dependent Operators: ROA

Mutation of speeds:
Small probability for each waypoint (gene) for a speed change.
Mutation of routes:
One mutation for each route.
Random selection of two indices of the waypoints of the route.
Removing the part of the route between the selected indices.
Random selection of a new waypoint and inserting the route from the
start point of the selection via the new waypoint to the end of the
selected part of route.
Problem Dependent Operators: ROA

Crossover of taxi routes:
Selection of a pair of chromosomes.
Selection of two indices for crossover.
Exchanging the parts of the routes between the selected indices.
Insert connections (taxi route) between the last point of the first
remaining part of original route to the start of the exchanged part and
from the last point of the exchanged part to the beginning of the
second remaining part.
Visualization of Crossover
Problem Dependent Operators: Other

Add holdings on waypoints:
Insertion of holdings in a taxi route in dependence of the success of
optimization.
Probability to be selected as holding point depends on the Safe Node of a
route (last safe holding position in case of conflicts) and the type of the
nodes (already holding point, stop bar exists for this node, no runway node).
Random selection of node.
Adapt speed changes:
In case of more than one speed change within a certain distance the second
speed change is moved behind the prescribed distance.
Example for Adaptation of Speed Changes

Speed correction for a minimum length of 400 meters
Distance for the speed correction marked by red lines.
35
30
25
km/h
20
15
10
100
200
300
400
500
600
Meter
Original Route
Speed Control
700
800
900
1000
Core Element: Conflict Detection
Position p on the link W:

= i + i (i + i ) ((i+1) i)
Minimum distance between two links U and W (Calculation of extreme values):

2
dis 2 = (wi , j u k , j + (v1t + v1t 2 )1 ( wi +1, j wi , j ) (v2 t + v2 t 2 )2 (u k +1, j u k , j )) 2

j =1
Monitoring and Adaption of Actual Position:

Possible deviation types:
All: Adaption to actual position

and speed.
Correct taxi route, no conflicts

and a negligible difference to
target time.
Nothing
Correct taxi route, no conflict

and the difference to target
time is to high.
New optimization for a

smaller difference to target
time.
Wrong route
Conflict
New route including

actual position.
New optimization.
the
Reality Check: 1st Workshop at DLR
Simulator
TRACC
Real Controller
Command
Panel
Controller
Display
Flight
Tables
Summary
Introduction of conflict-free time based taxi routes.
Problem solution done with evolutionary algorithms.
1st Workshop gave go for further development.
Thank you for your interest.
Questions?

EURO2013 Creation of Conflict-Free Taxi Routes For Aircraft With Evolutionary Algorithms

Uploaded by

Copyright:

Available Formats

EURO2013 Creation of Conflict-Free Taxi Routes For Aircraft With Evolutionary Algorithms

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

EURO2013 Creation of Conflict-Free Taxi Routes For Aircraft With Evolutionary Algorithms

Uploaded by

Copyright:

Available Formats

www.DLR.

Creation of Conflict-free Taxi Routes for

Taxi Routing on Airports: Todays Situation

Taxi Routing on Airports: Future Vision

TRACC-Approach: Creation of a Conflict-free and Optimized

Evolutionary Algorithm of TRACC

Problem Dependent Operators: TOA

Problem Dependent Operators: ROA

Problem Dependent Operators: ROA

Problem Dependent Operators: Other

Example for Adaptation of Speed Changes

Core Element: Conflict Detection

Position p on the link W:

Minimum distance between two links U and W (Calculation of extreme values):

dis 2 = (wi , j u k , j + (v1t + v1t 2 )1 ( wi +1, j wi , j ) (v2 t + v2 t 2 )2 (u k +1, j u k , j )) 2

Monitoring and Adaption of Actual Position:

All: Adaption to actual position

Correct taxi route, no conflicts

Correct taxi route, no conflict

New optimization for a

New route including

Reality Check: 1st Workshop at DLR

Thank you for your interest.

You might also like