Emotion Simulation For Virtual Human In Intelligent Virtual Environments

Affective computing is an emerging research field. It constructs a harmonious human-machine environment and improves the computer intelligence by enabling computer to recognize, understand,express and adapt to the user’s emotions. Affective computing can be divided into affect detection, emotion modeling and emotion expression. We lay our efforts mainly on emotion generation and expression of virtual human. In order to focus on emotion modeling, intelligent virtual environments should be constructed that it can shield off the physical details of virtual human construction and react to high-level virtual human interaction. We adopted semantic based methods to build this kind of intelligent virtual environments (IVEs).In virtual human emotion modeling fields, the current emotional models are basically ground on cognition so that they omit the affect of basic human emotions. In addition, the simulation of uncertain human emotions is still a challenge. We provided uncertain emotion generator under multi-stimulus to solve these problems. We used particles to sample the uncertain emotions and. modified these samples by cognition. Then human emotions can be represented by the weighted values of these particles. In this way, we combined basic human emotions and complex emotions together and constructed uncertain emotion process.Only emotion model is not enough to construct believable virtual human. Emotions must be presented outside through their bodies. We built facial expressions and body behaviors separately. In facial expression fields, we built six basic facial expressions and blended them according to the probability of basic emotions in the mixed emotional states. In human emotional behavior construction aspect, we divided the human actions into movement fragments of different body parts, then labeled these fragments manually to record the temporal and spatial constrains, at last we combined the fragments of different body parts to form multi-model emotion expression sequence.Emotion transmission in virtual crowds is our research focus. According to the features of crowds emotion propagation and for the sake of reducing the computing time, we proposed that heat conduction model can be used to simulate virtual crowds emotion transmission. We adopted alternating direction implicit (ADI) method to calculate the difference results and compute unified device architecture (CUDA) platform to accelerate the speed. We can run a real-time crowds simulation for thousand scale.