(slides) p-tour: a personal navigation system for tourist

31
P-Tour: A Personal Navigation System for Tourist Atsushi Maruyama Xanavi Informatics , Naoki Shibata Shiga University , Yoshihiro Murata Nara Institute of Sci. and Tech., Keiichi Yasumoto Nara Institute of Sci. and Tech., Minoru Ito Nara Institute of Sci. and Tech.

Upload: naoki-shibata

Post on 09-Jun-2015

3.237 views

Category:

Technology


4 download

DESCRIPTION

http://ito-lab.naist.jp/themes/pdffiles/041019.atsu-mar.ITSWC2004.pdf Maruyama, A., Shibata, N., Murata, Y., Yasumoto, K. and Ito, M.: P-Tour: A Personal Navigation System for Tourism, Proceedings of 11th World Congress on ITS Nagoya, pp.18-21 (October 2004) We propose a personal navigation system for tourism called P-Tour. When a tourist specifies multiple destinations with relative importance and restrictions on arrival/staying time, P-Tour computes the nearly best schedule to visit part of those destinations. In addition to the map-based navigation, P-Tour provides temporal guidance according to the schedule, and automatically modifies the schedule when detecting the situation that the tourist cannot follow the schedule. We have developed a route search engine as a Java Servlet which can compute a semi-optimal schedule in reasonable time using techniques of genetic algorithms.

TRANSCRIPT

Page 1: (Slides) P-Tour: A Personal Navigation System for Tourist

P-Tour: A Personal Navigation System for Tourist

Atsushi Maruyama Xanavi Informatics ,

Naoki Shibata Shiga University ,Yoshihiro Murata Nara Institute of Sci. and Tech.,Keiichi Yasumoto Nara Institute of Sci. and Tech.,Minoru Ito Nara Institute of Sci. and Tech.

Page 2: (Slides) P-Tour: A Personal Navigation System for Tourist

Outline of our presentation

1. Background2. Overview of P-Tour3. The route search engine4. Evaluation5. Conclusion6. Future works

Page 3: (Slides) P-Tour: A Personal Navigation System for Tourist

Background(1) High performance PDA Small built-in GPS unit on mobile phone Wireless LAN, 3G mobile phone, PHS

Navigation system on mobile phone or PDA

Navigation service on mobile phone : EZ Navi Walk on mobile phone by au kddi

• Built-in GPS unit on mobile phones• Route search between two locations• Guidance by voice, text, etc.

Page 4: (Slides) P-Tour: A Personal Navigation System for Tourist

Background(2)

Existing navigation systems– Car navigation system– Personal navigation service by au kddi– Etc.

Functions are limited

–Route guidance between two locations

Inadequate for tour navigation

Page 5: (Slides) P-Tour: A Personal Navigation System for Tourist

Background(3) Navigation system for tour navigation

•Many destinations to visit

•Each destination may have business hours, appointed time, etc.

•User may wish visiting destinations as many as possible

•If it is impossible to visit all destinations, system should choose part of destinations to visit

•Importance value can be specified for each destination•Timezone of visits can be specified for each destination•Guidance function includes time schedule management

•Many destinations to visit

•If it is impossible to visit all destinations, system should choose part of destinations to visit

•User may wish visiting destinations as many as possible

•Each destination may have business hours, appointed time, etc.

We propose personal navigation system with these features

Page 6: (Slides) P-Tour: A Personal Navigation System for Tourist

P-Tour : A Personal Navigation System for Tourism

Tour scheduling

•Beginning and ending locations of the tour

•Importance of each destination

– User inputs: •All destinations and corresponding timezones

Page 7: (Slides) P-Tour: A Personal Navigation System for Tourist

P-Tour : A Personal Navigation System for Tourism

Tour scheduling– Pre-calculation is performed within 10 sec.

– System outputs:• Route with arrival/departure time for each

destination

Page 8: (Slides) P-Tour: A Personal Navigation System for Tourist

P-Tour : A Personal Navigation System for Tourism

Tour scheduling

Request

Schedule

Request

Schedule

User Server

Incremental tour scheduling• Adding new destinations• Changing importance of destinations One step of incremental calculation is performed in a

few seconds

Page 9: (Slides) P-Tour: A Personal Navigation System for Tourist

System overview of P-Tour

Client (cell phone or PDA)•Route guidance program•User interface

Implemented with Java MIDlet

Internet

ServerRoute search engineImplemented as a Java servlet

Map database

Destinationdatabase

RequestSchedule

Page 10: (Slides) P-Tour: A Personal Navigation System for Tourist

Route guidance mode

Schedule display

The entire route

Moving along the scheduled route

When visiting a destination.

Remaining stay time/Departure time

Schedule display

Arrival/Departureand stay time

Page 11: (Slides) P-Tour: A Personal Navigation System for Tourist

Automatic recalculation of schedule

When user goes into wrong route When user’s moving speed is too slow When user stays at a destination too long

These situations are automaticallydetected using GPS and clock

The system warns user Displays a route to return to the original route Changes the schedule and route

Page 12: (Slides) P-Tour: A Personal Navigation System for Tourist

Requirements for the route search engine

Fast enough for the incremental scheduling– 10 seconds for the initial calculation– A few seconds for a recalculation

The output route should maximize user’s satisfaction– Fitness function converts route to

satisfaction rate

Page 13: (Slides) P-Tour: A Personal Navigation System for Tourist

Fitness function Output route should include important

destinations as many as possible

Each destination should satisfy a corresponding time restriction

Output route should an efficient route without detours

Importance values of included destinations are added to the fitness value

Importance values are only added if each of destinations satisfies the restriction

Total distance of user’s movement is subtracted from the fitness value

Page 14: (Slides) P-Tour: A Personal Navigation System for Tourist

Fitness functionNumerical expression of the fitness function

otherwise

nsrestrictio its satisfiesand route in the included is if

nsdestinatio all

0

1)(

of valueImportance)(

)movement of distance total(

)()(

valueFitness

d

d

dS

ddI

dIdS

is a constant

Page 15: (Slides) P-Tour: A Personal Navigation System for Tourist

Fitness functionSetting of value and output route

Value of and the output route

low medium high

Low value leads to detourHigh value leads to destinations near to the beginning location of the tour to be only selected

It is desirable to set value according to user’s preference

Page 16: (Slides) P-Tour: A Personal Navigation System for Tourist

Route search algorithm Route between all combinations of

two destinations are calculated– A* algorithm is used

• In our experiment, moving speed is assumed to be 30km/h for usual road, 60km/h for express ways

– same as car navigation system• Moving speed can be obtained from map or

other data sources– Routes between known destinations are

calculated beforehand• Routes to/from newly entered destinations

are calculated extempore

Page 17: (Slides) P-Tour: A Personal Navigation System for Tourist

Route search algorithm

Determining visiting order of each destination by genetic algorithm

Advantages of using GA

GA always retains multiple candidate solutions

•It is always possible to return approximate solutions

•User can choose preferred solution from them

GA is used to obtain approximate solution for combinatorial optimization problem

Page 18: (Slides) P-Tour: A Personal Navigation System for Tourist

Overview of Genetic Algorithm

Candidate solutions are generated randomly

NationalMuseum

法隆寺HoryujiTodaiji Kofukuji Yakusiji

Beg

innin

glo

catio

n

En

din

glo

catio

n

Candidate solution

GA always retains multiple candidate solutions

Page 19: (Slides) P-Tour: A Personal Navigation System for Tourist

Randomly selects two solutions, and make a new solution from themCandidate

solutions for the next iteration

Calculate fitness values and select solutions with relatively high fitness values

Repeat the iteration until predefined iteration count expires

Page 20: (Slides) P-Tour: A Personal Navigation System for Tourist

Evaluation of our system Equipments/settings

– Server HW: A personal computer with Pentium4 2.4GHz– Server SW: Linux(Debian), Java Servlet, Tomcat 4.2– Map data format: Navigation System Researchers’ association digital r

oadmap format– Navigation area: North Nara– Moving method : by car– GA iterations(generations) : 100– Settings of constants : γ=1

Things evaluated– Validity of output route– Time to calculate routes– Difference between optimal and output solutions

Page 21: (Slides) P-Tour: A Personal Navigation System for Tourist

Calculation time of route

Calculation time of routes between any combination of two destinations are not included

•Converges at 50th iteration•Outputs a satisfactory route within 10 seconds

Fitnessvalue

Computation time

Number ofiterations

Page 22: (Slides) P-Tour: A Personal Navigation System for Tourist

Difference between optimal and output solutions

Difference is about 1% Sufficient for practical use

NumberOf dest.

Fitnessvalue

Optimalvalue

Difference(%)

Calc. Time(P-Tour)

Calc. Time(Optimal)

Page 23: (Slides) P-Tour: A Personal Navigation System for Tourist

Enhancement:multi-objective scheduling

Minimization of multiple fitness functions– Traveling cost– Time

Osaka

Kyoto

Express

Special Express

Fast, but expensive

Airplane + train

Slow and expensive

Slow, but cheap

Find a set of routes which are worth consideration

Page 24: (Slides) P-Tour: A Personal Navigation System for Tourist

• User can choose one schedule from several candidate plans• Actual routes are more intuitive than set of values

Enhancement:multi-objective scheduling

Satisfaction : 119Cost : 0

Satisfaction : 154Cost : 80

Satisfaction : 182Cost : 2520

Page 25: (Slides) P-Tour: A Personal Navigation System for Tourist

Conclusions

We proposed P-tour– Tour scheduling using GA– Timezones can be specified

We evaluated P-tour– Search time is about 10 seconds

Page 26: (Slides) P-Tour: A Personal Navigation System for Tourist

Ongoing works Supporting tour using multiple

transportation methods – Car, train, bus, walking, etc.– Appropriate route can be selected using

multi-objective scheduling Improvement of user interface

The route automatically change when context changes• When it begins to rain, user may want to visit

indoor exhibition• Group of users can break up and get together

using P-Tour

Page 27: (Slides) P-Tour: A Personal Navigation System for Tourist

Thank you.

Page 28: (Slides) P-Tour: A Personal Navigation System for Tourist

Overview of route search algorithm

•Destination ID•Wait time•Stay time•…

GA always retains multiple candidate solutions Candidate solutions are encoded

Dest. 1 Dest. 2 Dest. 3 Dest. n…

Page 29: (Slides) P-Tour: A Personal Navigation System for Tourist

User’s input

Beginning and ending locations of the tour:   NAIST

Tour begins at:   9:00amTour ends at:   9:00pm

No. Name Importance Arrival Stay

D1 Toshodaiji Tmpl. 5 - 60minD2 Yakusiji Tmpl. 5 - 60minD3 Horyuji Tmpl. 5 <=15:00 >=180minD4 Fujinoki Tomb 5 - 90minD5 Heijo-kyo 5 - 150minD6 Saidaiji Tmpl. 5 - 30minD7 Gakuem-mae 5 <=19:30 >=60minD8 Sarusawa pond 5 - 60minD9 Botanical garden 5 - 45minD10 National Museum 5 - 60minD11 Shin-Yakusiji Tmpl. 5 - 60minD12 Shou-sou-in 5 - 30minD13 Tai-an-ji Tmpl. 5 - 60minD14 Hou-rin-ji Tmpl. 1 - 60min… … … …D30 Hokke-ji Tmpl 1 - 30min

Page 30: (Slides) P-Tour: A Personal Navigation System for Tourist

Validity of output route

D3

D1D2

D7 D12D9

D8

D10

D1 ~ D13 Importance 5D14 ~ D30 Importance 1Timezone D3 ≦15:00 D7 ≦19:30

Destinations in outputD1, D2, D3, D7, D8, D9D10, D12

Arrival timeD3 14:50D7 19:10

INPUT

D4

D13

We changed importance to 10, and recalculated the route

Page 31: (Slides) P-Tour: A Personal Navigation System for Tourist

Validity of output route

D3

D1D2

D7 D12D9

D8

D10

D1D2

D3D4

D13

D5D7

Before After D4,D13 Importance 10