Douglas Crossley
Alexis Oyama
Douglas Crossley
External Advisor:
Kostas Bekris
Project Introduction:
When designing a building, Architects take into consideration many things from what types of materials to increasing capacity and one thing all Architects keep in mind is how to protect their building from the one threat that is the most preventable, and yet most common disaster, fire. Along with making sure that the flammable materials are well modulated and sprinklers are properly placed, they must also take into consideration the safety of the potential of people that will be occupying the space. These people must be accounted for in emergency situations of all conditions (i.e. max/min capacity, building knowledge etc.). This reasoning provides the basis for our software, as we believe with modern computer graphics, aided with realistic Artificial Intelligence, we can realistically simulate the evacuation scenario of a building containing a variable amount of occupants in order to understand and optimize the building ‘s design in terms of fire safety.
The overall goal is to have a software that will be able to realistically simulate the evacuation process and to maximize the risk reduction. Though, simple models do exist for moving Artificial Intelligence agents toward a target destination while avoiding fire hazards, it is an all together different task to realistically simulating a genuine human’s response to such an event. There are many factors including building knowledge, fire safety training and even panic that contribute to the decisions a human will make. We would like our software to take advantage of a more comprehensive Intelligence system that would allow the user to factor in variables about the occupants as well as ones about the building design that would ultimately simulate a more realistic scenario. In our program, we have decided at this point, and for demonstration purposes, to include one or two predesigned buildings to test the simulation. We are thinking that it could be interesting to test buildings that already exist on campus like SEM or AB. At some point, we would like to make it possible to allow the user to custom design their building according to specific realistic design needs.
To achieve this, our system will utilize the PRACSYS open source framework to simulate a virtual environment with multiple agents using different level of intelligence; therefore, allowing the user to determine how the design a building is effective in terms of safety and how training people will result in an improvement in speed and efficiency of exiting a building. By simulating such an emergency event, building planners and engineers could effectively design a building to anticipate realistic human reaction in an emergency scenario.
By integrating such a framework with another open source Graphical User Interface system, our program will achieve the perfect integration of built in functionality and intelligence with a sleek user interface design. Allowing the user to view the simulation from a variety of pre-set and user-defined camera angles, while giving the user the ability to generate fire and manipulate the simulation agents in real-time. These simulations will process correlated data such as integrity of building, time to escape, injuries etc. in order to test existing fire capacity regulations and for setting new fire safety standards for future buildings and their occupants.
Simplified Class Structure:
Application Interface:
