Project Report 2 : Control the map
Project Report 2 : Control the map
Student :
Olivier Richard
Aurelien Bonnafont
What has change
Our initial project was to use face recognition to allow a user to “kill” an opponent. We do not think that we are going to implement this feature because we do not have access to the screen image and then, we cannot send it to the server. We have decided to use an design. Now, each user is linked to a vufodia marker. An user will detect those vufodia marker and try to “kill” a player. That suppose user has wearing a t-shirt with a such marker and are handling a paper with the marker on. This is unrealistic and not really good for the user experience but as far as we know, this is the only way to make a playable game.
Advancement
The server side part is now finished. We have decided to use Google app engine because it is an easy way to deploy a Java server. We have decided to not use a database and only store the data has objects. We do not intend to implement history or ladder, so we do not think the time spend in writing the queries is worth it.
We have also decided to simplify the architecture. Initially, we expected to use bi-dimensional connection (mostly socket) but it happened that socket are not working well with jetty, the Java server deployed on Google App Engine. We have tried the Xmpp protocol but it requires that each user has a registered account. Even if Xmpp works well for our purpose, we do not think user are ready to create an XMPP account (on another service than our) to play so we decide to not continue in that way. Instead, we are going to make repeated calls of the same function using window.setInterval in javascript. This is not the best solution in term of CPU usage for both side but we assume that :
– the difference for the client is negligible,
– the server can handle the amount a connection regarding that there will not be more than 20 or 40 players.
In summary, we try to improve the user experience rather than the scalability.
We have also designed a solution to display users and control points on the map using Google Map. We are currently search a way to add 3D building in the map to help the user in its orientation. Because Argon is still crashing a lot, we do not have to try to implement VE content in Argon Page right now.
What is next
Implement the ArgonPage. This should not be too complicated because the back-end is already made. We just have to call the right function and then create the appropriate 3D content.
In the first chapter, we have said that we have switch the face recognition with vufodia marker recognition. We have not decide yet the implementation in the server side. Indeed, to still let an unknown part, we would like to let Argon search for more marker than there is player. In that way, there is still a challenge and a chance to miss. As soon as the new version of Argon will be available, we will experiment different solution to understand the load that each marker cause.
Finnaly, we have to create a css designed for mobile device.
The current service is available at :
http://argonolivier.appspot.com/
(We try to keep it functional but our wifi network at home does not allow us to connect on our computer from iphone or ipad using that networks so we often ship new code on that server. So, the version deployed should be working but we cannot offer an availability of 100%)
Comments are closed.