Weekly Report 51

Engineering:

Worked Extensively with DASSIE stuff

Closed Tickets :  447, 443, 444, 441, 438

Implmented new map integration into Behavior Shop

Generated new SPSS map for DASSIE

Extended DASSIE to allow for human guided decomposition

Implemented Decomp visualization in FI3RST

Fixed portions of SPSS python code

Sci:

Planned out paper for FDG, worked on Research plan for FDG doctorial event.

FDG Paper Outline

Introduction:

Introduce the concept of level decomposition and other methods of  providing information to Agents.  Talk briefly about how many different types of implementations there are for each of these.  Make the point that for navigation there are just good and bad methods with no clear winners.  Point out the uses for decomps and other non point based nav meshes for things beyond navigation.  Then go into the need to differentiate  between quality of decomps for things beyond navigation.  Talk about and preview related work here about different metrics for decomps.  Introduce quickly how I will be presenting metrics.  Discuss the advantages of being able to choose the “best” decomp without using the word best.  Present it more as optimizing for your particular application.

Related Work:

Present the usual crew of suspects for other full world decomposition metrics here

Methodology

Present my contribution of 4 decomp metrics here.  Produce a world and apply multiple hand generated decomp methods to it (3/4 methods – SFV, ASFV * 2, HM, Delanay) use this world as a running example to demonstriaght each metric.  After explaining each metric show the comparision graphs they would generate for each method.

Experimentation

Show some metrics in use for a world or hold off on showing the final reports I discussed above and list them for several worlds here.

Conclusion

Having metrics to sort the generated results from the different decomp methods is a great thing

Comments are closed.