Human Motion Planning in Spatial Audio Field

Student: Xu, Shuyang (UID: 3035947740)

Supervisor: Prof. Komura, Taku

Character animation is a specialized area of the animation process, which involves bringing animated characters to life.

–Wikipeida

Obejectives

Human Motion Synthesis

Design a model which can synthesize human motions effectively given a spatial audio field. The human motions are expected to be different when the location of the audio source is different.

Detaset Establishment

Establish a new dataset which is suitable for training a model used for human motion synthesis in a spatial audio field. The dataset is epected to be collected using a inertia mocap system.

Baseline

Fill in the research gap in the area of 3D human motion synthesis and put forward new metrics as indicators to the fidelity of the generated human motions in a spatial audio field.

Closely Related Works

Motion Diffusion Model

A model that can synthesize high quality human motions given text descriptions using the diffusion process.

Tourist taking photo of a building
Windows of a building in Nuremberg, Germany

Bailando

A model that innovatively combines Vector Quantized-Variational AutoEncoder
(VQ-VAE) with Generative Pretrained Transformer (GPT) and can synthesize human motions with high fidelity given audio.