Enhancement of collision avoidance technique in dynamic crowd simulations using visual cues and vision based algorithm

Iznora Aini Zolkifly (2022) Enhancement of collision avoidance technique in dynamic crowd simulations using visual cues and vision based algorithm. Doctoral thesis, Universiti Malaysia Sabah.

	Text 24 PAGES.pdf Download (604kB)
	Text FULLTEXT.pdf Restricted to Registered users only Download (3MB)

Abstract

Crowd simulations are widely used in virtual environment applications such as in entertainment, games, military training, safety issues, evacuation, and emergent situation. It visualizes the crowd behaviors, movement, and interactions in simulating a virtual environment. A dynamic crowd's density can be classified as high, medium or low depending on the situation. The abrupt obstacles occurrences may emerge at any time and anyway in a dynamic crowd simulation. At the same time, the dynamic changes of the obstacles would initiate a complicated path to obtain a prediction of the target’s motion. Due to a lack of contacts amongst agents, they will interact only when prompted, such as in identified close proximity circumstances. In reality, each agent has its own goal and personal space. However, in an emergent or panic situation, the agent’s goal can be altered to find the optimal free collision path to exit the environment. Therefore, agents could perceive the environment and try to find the optimal solutions with the aid of visual cues. The main objective of this study is to produce an enhanced technique that integrates the visual cue outcome in the collision avoidance pipeline during emergent situations. The proposed technique emphasised the need of avoiding sudden obstacles by including the bounding circle mechanism and a vision-based method into the free-path environment in order to achieve optimal and better exploration in crowded conditions. From the experiments conducted, the suggested method can reduce all evacuation time effectively and improve the overall proposed collision avoidance and pathfinding techniques in a dynamic crowd simulation environment. The Predetchenky and Milinskii method has been carried out to validate simulation results obtained by the Netlogo testing agent's evacuation system. The findings of this study can be used to mimic the evacuation of a single-layer floor of a structure, such as computer labs, science labs, hallways, or schools. The simulation can provide vital information for evaluating a design, assisting in the planning process, and training to cope with emergencies.

Item Type:	Thesis (Doctoral)
Keyword:	Crowd simulations, Virtual environment applications, Algorithm
Subjects:	Q Science > QA Mathematics > QA1-939 Mathematics > QA1-43 General
Department:	FACULTY > Faculty of Science and Natural Resources
Depositing User:	DG MASNIAH AHMAD -
Date Deposited:	13 Sep 2024 09:33
Last Modified:	13 Sep 2024 09:33
URI:	https://eprints.ums.edu.my/id/eprint/40507

Actions (login required)

View Item