James' Scratch Simulations
Gorillas Remake
Paw Patrol Path Game
Gorillas Remake
- Uses parallel computing to draw the city scene so quickly
- Right now both gorillas are controlled by computer and hey take turns throwing random angles and velocities.
- Placeholder for AI project for gorillas to learn how to aim better
Tanks Project
Paw Patrol Path Game
Gorillas Remake
This was made from the ground up to be an AI project. Plan was to do genetic algorithm that feeds into a neural network with inputs based on what the tanks are sensing.
Right now the tanks just move randomly, but the sense mechanics are there, and it's basically primed/ready start start the AI part.. :)
Paw Patrol Path Game
Paw Patrol Path Game
Paw Patrol Path Game
Remake of a 'board game' that we bought from Dollar General. I scanned the game board and pieces for graphics
I wanted to see know on average how many turns a game lasts, and to see if starting turn order makes any impact on winner - results below
Paw Patrol Path Game: Simulation Findings
It seems in all cases, it's better to go first in this game, most notably for 2-player games.
The average game is 19 turns (moves for each player), or roughly 9-12 min per player
4 player games tend to be shorter, which was not something I anticipated. I suspect this is due to more players being able to block the bad action board squares.
James' Other Simulation Projects
Pathfinding Sandbox
Pathfinding Sandbox
Pathfinding Sandbox
I wrote this program for my graph theory (math class) final project in 2014. It was written in Fantom programming language.
Traffic Simulation
Pathfinding Sandbox
Pathfinding Sandbox
I simulated my daily (in-town) commute through Stockton California to school for my simulations class final project, to explore the cost of driving faster and advantage of looking further ahead.
Simulation Labs
Pathfinding Sandbox
Simulation Labs
We used Vensim and other technologies to perform simulations of animal breeding, line queues, employee demand, and more.
Pathfinding Sandbox
Traffic Simulation Details
Alpine road, Stockton California
My partner, Bri Prebilic Cole, and I simulated traffic for ~2km distance from Pacific Avenue to West Lane, which has 4 stoplights and 1 pedestrian crossing.
Data Collection
We collected data for 3 cycles of lights at each of these stop lights and used the averaged values shown to the right for our simulation.
We counted the pedestrians we saw while we were timing spotlights, and averaged that over the time we were there recording to get about one
pedestrian every minute and a half.
Driver Update Rules
Each vehicle records and reports data on its gas consumed, brakes consumed, velocity, and position in the system. Gas consumed is measured by the accumulation (or integral) of a vehicles positive acceleration, while brakes consumed is the integral of a vehicles negative acceleration. Constant velocity, or zero acceleration, is considered “free” in our system and counts towards neither brakes or gas.
Two Driver Profiles
Vehicles are the agents that move through our simulation. Each vehicle is spawned at distance zero, and assigned a random lane (lane 1 or lane 2) and a random driver type (fast driver or slow driver), with 50%
probability of each.
We used real data for the max speeds, looking at two speed
radar signs along Alpine Ave. We collected several speeds for a span of about a
half an hour. We ordered the speeds and split the data in half, and took the
average of the two halves to get the speeds of the slow and fast cars.
Assumptions/Limitations
- Cars only starting from either side of the 'track', not entering from any of the intersecting streets
- Pedestrians cross road in 5 seconds
- Friction-less Roads
- Light timers are uniform
- Only two types of drivers - fast and slow drivers
Technologies Used
Microsoft Visual Studio 2012
Microsoft Team Foundation Version Control
Microsoft Library: System.Windows.Forms.DataVisualization.Charting