Teófilo Bezerra DutraBlack River Studios (Samsung), Manaus, Brazil
Skills and Expertise
Mar 2011 - Jun 2015
- PhD Student
- Research focused on Crowd Simulation. Developed a new model for simulating crowds based on synthetic vision. The point of view of each agent is rendered (to a texture) and evaluated in order to determine the best way of adapting its motion.
Mar 2009 - Feb 2011
- Master's Student
- Research focused on Crowd Simulation. Developed a hybrid model for simulating crowds combining approaches based on potential fields and social forces.
Research Items (13)
Exploring recent developments in the rapidly evolving field of game real-time rendering, GPU Zen assembles a high-quality collection of cutting-edge contributions for programming the GPU. GPU Zen includes coverage of the areas of geometry manipulation, lighting, general rendering, screen-space techniques, Virtual Reality and general compute tasks.
Most recent crowd simulation algorithms equip agents with a synthetic vision component for steering. They offer promising perspectives through a more realistic simulation of the way humans navigate according to their perception of the surrounding environment. In this paper, we propose a new perception/motion loop to steering agents along collision free trajectories that significantly improves the quality of vision-based crowd simulators. In contrast with solutions where agents avoid collisions in a purely reactive (binary) way, we suggest exploring the full range of possible adaptations and retaining the locally optimal one. To this end, we introduce a cost function, based on perceptual variables, which estimates an agent’s situation considering both the risks of future collision and a desired destination. We then compute the partial derivatives of that function with respect to all possible motion adaptations. The agent then adapts its motion by following the gradient. This paper has thus two main contributions: the definition of a general purpose control scheme for steering synthetic vision-based agents; and the proposition of cost functions for evaluating the perceived danger of the current situation. We demonstrate improvements in several cases.
Crowd Simulation is very important in many virtual reality applications, because it improves the sense of immersion of the users by making the population of agents in the environment to move as real crowds do. Recently, models for simulating crowds, in which each agent is equipped with a synthetic vision system, have shown interesting results regarding the natural manner in which the agents navigate inside the environment thanks to their visual perception. In this article, we propose an upgrade to the agent's visual system with a panoramic view in order to allow an agent to expand its vision beyond the limit of $180^\circ$ imposed by the common projection provided by rendering APIs. Also, we analyze different parameters, which are used to define the field of view, to investigate the influence they have on the agent's behavior. The impacts that those changes may cause on the efficiency of the algorithms are also analysed. A visible change on the agent's behavior is achieved by using the technique, with a slight loss of performance.
Crowd Simulation is very important in many virtual reality applications, because it improves the sense of immersion of the users by making the population of agents in the environment to move as real crowds do. Recently, a model for simulating crowds, in which each agent is equipped with a synthetic vision system, has shown interesting results regarding the natural manner in which the agents navigate inside the environment thanks to their visual perception. In this article, we propose an upgrade to the agent's visual system with a panoramic view in order to allow an agent to expand its vision beyond the limit of 180 • imposed by the common projection provided by rendering APIs. Also, we analyze different parameters, which are used to define the field of view, to investigate the influence they have on the agent's behaviour. The impacts the changes may cause on the efficiency of the algorithm are also analysed. A visible change on the agent behaviour is achieved by using the technique, with slightly loss of performance.
Abstract Following is a very common type of interaction between individuals observed in crowds. For example, in streets or in corridors, following occurs as soon as one's motion is constrained in the sagittal plane, while overtaking possibilities are constrained as well by walls or people coming in counter flow. In this paper, we present a new model to simulate following behaviors. We explore the distance at which followers try to match the leaders’ speed. In our model, the following distance is dynamic and changes according to the follower's speed and the prediction of the leader's future position. We evaluate the model using real data and we also compare it with other following models showing the improvements.
In this paper, we present a new model to simulate following behavior. This model is based on a dynamic following distance that changes according to the follower’s speed and to the leader’s motion. The following distance is associated with a prediction of the leader’s future position to give a following ideal position. We show the resulting following trajectory and detail the importance of the distance variation in different situations. The model is evaluated using real data. We demonstrate the capacity of our model to reproduce macroscopic patterns and show that it is also able to synthesize trajectories similar to real ones. Finally, we compare our results with other following models and point out the improvements.
Computer simulation of realistic crowd behavior has been the focus of active research for more than two decades now. In crowd simulation, there is usually a trade-off between performance and realistic crowd behavior. In this paper, we propose a model, based on potential fields, that enables the introduction of many behaviors in crowd simulations, while keeping good performance. The model uses multiple groups to guide agents to various different goals in the environment, and combines potential fields and reciprocal velocity obstacles (RVO) approaches, where the first sets the preferred velocities of the agents according to their current goals, whereas the second makes the agents avoid collisions. We used three scenarios to demonstrate the capabilities of our model for simulating crowds in which the agents present greater variety of behaviors in real-time without using a complex architecture.
In this work, we propose a computer model for simulating and studying crowd evacuation behavior, and apply it to study the evacuation of a soccer stadium. Our model uses 2D cellular automata defined over multiple grids that represent different levels of the simulated environment. A pedestrian moves from a given grid cell to another by analyzing multiple movement options and resolving conflicts with other pedestrians. Validation tests were performed, demonstrating that the proposed model is able to simulate crowd evacuation successfully.
Environments populated with crowds are employed in various applications, such as games, simulators and editors. Many of these environments require not only a realistic and detailed rendering, but it must run smoothly in real-time. This task easily exhausts the system's resources, even considering the current state-of-the-art hardware. Therefore, crowd rendering in real-time remains a challenge in computer graphics. Approaches exploiting levels of detail, visibility culling and image-based rendering are presented in the literature to facilitate this task. The first two increase the efficiency of rendering, but sometimes are not enough to keep an interactive frame rate. Some researches on this subject focus on image-based rendering techniques, specifically with the use of impostors. In this work it is proposed a method that balances the computational demand of rendering job by varying the threshold's distance of the representation switch between full geometry (mesh) and image-based (impostors) models in accordance with the available resources.
Crowd simulation is a computationally expensive task. Many models to simulate crowds have been developed over the years and can be classified into two big groups (macroscopic and microscopic) according to how the agents are managed. In some macroscopic models the agents are grouped and guided by the potential field of their group. The construction of potential fields is the bottleneck of those models, so it is necessary to use a small number of groups in order for a simulation to run at interactive frame rates. In this article, it is proposed a model based on a macroscopic model, which, mainly, aims at reducing the cost of computing the potential fields of the groups, by using groups discretized according to the needs of the environment. At the same time, it is proposed the addition of groups that can steer the agents of a simulation to momentary goals, which gives the crowd a wider variety of behaviors. Finally, it is proposed the use of a social forces model to prevent collisions between agents and between agents and obstacles.