[
https://issues.apache.org/jira/browse/SIS-325?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hugo Rivera updated SIS-325:
----------------------------
Attachment: proposal.6.pdf
Another draft of the proposal. This mentions multiple route planning.
> [GSoC] Optimizing patrol routes with multiple patrol cars
> ---------------------------------------------------------
>
> Key: SIS-325
> URL: https://issues.apache.org/jira/browse/SIS-325
> Project: Spatial Information Systems
> Issue Type: New Feature
> Reporter: Suresh Marru
> Attachments: proposal.5.pdf, proposal.6.pdf
>
>
> The proposed project extends the early effort described at
> http://www2.tcs.ifi.lmu.de/~ramyaa/publications/patrol.pdf
> As described in the paper, in early efforts a single patrol car was assumed
> to derive patrol routes based on Voronoi diagrams. However the problem
> becomes more complicated when we have multiple patrol cars and multiple
> criminals. Some issues include: Do we spend multiple cars into crime areas
> with more crime rates? If one car is answering to a 911 call, then do we
> redistribute the other cars to cover the entire area? Consider a junction
> such as a T-junction, and let there be two cars â one on each arm of the âTâ;
> We would want to schedule the two cars such that they are not at the
> junction at the same time (i.e., when the car on the horizontal arm is in the
> middle, we want the car on the vertical arm to be at the bottom).
> Such patrolling and scheduling can be optimized using techniques such as
> genetic algorithms. In this, each patrol and schedule combination is
> assigned a value calculated by an objective function. Typically this is the
> average time taken by cars to answer 911 calls. We then use optimizing
> techniques to reduced this objective function.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
