Moose Habitat SurveyorFinal Presentation

University of Alaska AnchorageCS 401 : Software EngineeringCold Moose Chasers (Cold MC’s)

Max KonovalovChris Kulhanek

Jim Weller

Problem Statement

• Agencies like the forest service need a way to task biologists to measure and predict the amount of Moose edible biomass in an area

• Take random samples with a computer and project the distribution onto an area.

Scenarios

• Biologists get a list of GPS waypoints from an Agency like the forest service

• Biologists go to each Waypoint and sample 9 CurrentYearGrowth (CYG) of many PlantPatches

• Biologists export the biomass data to a file and give it to the Agency

30 meters

PlantPatch1

PlantPatch4

PlantPatch2PlantPatch3

Waypoint

5 meters

SurveyPole

SurveyPole

Foliage

Survey Pole

Survey Pole

FieldBiologist

Diameter of Patch

WayPoint

Survey Pole

Survey Pole

FieldBiologist

Diameter of Patch

Random Point on Diameter

3drp Pole

3drp Pole

0-1 meter

1-2 meters

2-3 meters

CYG

Random 3D Point

Legend

1 Map Base 3D Point2 Map Tip 3D Point3 Measure Base Diameter

CYG1

2

3

Features

• GUI Application– 3d Renderer– Model view– Data view of instance objects– Menus for

• Collaboration: Import/Export• Persistence: Save/Load

• Hardware– 3d Radio Pen (3drp)– Calipers

Architecture

• MVC– Model is well defined– Controller manipulates model

• Win32 Application

– Views render model• 3d Panel to show plant patches• TreeView control to show the model hierarchy• DataGrid control to show instance variables of

model components

Design

• Façade Pattern (FieldTrip)– Wraps Model class hierarchy– Methods with integers

• One Waywp = ft.getWp(wp_idx);pp = wp.getPp(pp_idx);cyg = pp.getCyg(cyg_idx);cyg.setCygBase(base3d)

• Façade Wayft.cygSetBase( wp_idx, pp_idx, cyg_Idx, base3d )

Design (con’t)

• Façade– Advantages

• Simple and fast to implement because you can use protected variables instead of set/get

• Less function calls to manipulate model• Model innards can be changed without affecting

interface

– Disadvantages• Not great OO design• Model components not very reusable

Design (con’t)

• Bridge– Data inputs should be able to use ink,

keyboard, or attached hardware– Provides flexibility in the field

Win32 Application

Viewmenus, folding tree,

directx picturebox, buttons, text boxes, property sheets,

dialogues, wizards, icons

ControllerEvent driven routines that

manipulate the model

Model

WaypointFieldTrip

1 *

PlantPatch

1 *

CurrentYearsGrowth

1 9

3DRP

LaserRangeFinder

Calipers

+getDistance()

Long2dMeasure

+grab3dSample()

3dMapper

+getDistance()

Short2dMeasurer

11

1

11

1Short2dTextBox

3dTextControl

Long2dTextBox

Expertise

• Chris– Built 3D meshes because of architectural

background– PlantPatch Model component– DataGrid GUI View Component

• Max– C++ string parser research– Waypoint Model component– TreeView GUI View component

Expertise (con’t)

• Jim– Project Lead– .NET C++ research– CurrentYearGrowth Model component– 3D PictureBox GUI View component– Secretarial slave (coffee, lunch, notes etc.)

Finished(?) Product

• We didn’t know until 5 minutes ago what exactly we were going to show you all

• Here it is…– Jim: Meshes– Chris: DataGrid– Max: TreeView from model file

Lessons Learned: Facts

• 50 files (70 expected)– 5 megs of source– 2500 Lines of code (3500 expected)– 85 pages of docs (120 expected)– 4 APIs (3drp, DirectX, TabletPC, .NET)

• 103 CVS commits– 51 CVS commits for Model.h

• 489 Total person hours (519 expected)– First 6 Weeks

• 3 hours of meeting time per week• 2 hours of research/design/writing time per week

– Last 8 Weeks• 10 hours of group time per week• 4-8 hours of code time per week

Lessons: Things we did right

• CVS is good, viewcvs makes it better, statcvs makes it even better

• C++ or not, Visual Studio is sweet

• Consistent meeting agendas and minutes increased traceability

• Good division of labor

• Long 8 hour work together meetings

• Code walk throughs

Lessons: Things that sucked

• Too many new technologies (Cray quote)• Lack of experience in C++ and C++ .NET especially• No client buy in. Should have had a contract • Prototype earlier, especially with new technologies• Specify expected application behavior better in scenarios• Old Libs and new libs (VC2003 vs. VC6)• Develop project schedule earlier and revise continually• Probably won’t achieve acceptance of original

requirements

Conclusions

We should have majored in

English

Just kidding…

Conclusions (for real)

• Software engineering is challenging

• 3 members does not equal 3 times the effort of a single member

• Good planning can save time and effort downstream

• Discipline is more important than genius

• Conservative is better than exciting

• Communication is the most important

THE END