Phase-Functioned Neural Networks for Character Control

Watch on YouTube

Show annotations

Download is disabled.

963,879

23,434

153

Genre: Film & Animation

Family friendly? Yes

Wilson score: 0.9924

Rating: 4.9741 / 5

Engagement: 2.45%

Yoshiboy2

Subscribe | 3.87K

Shared April 30, 2017

We present a real-time character control mechanism using a novel neural network architecture called a Phase-Functioned Neural Network. In this network structure, the weights are computed via a cyclic function which uses the phase as an input. Along with the phase, our system takes as input user controls, the previous state of the character, the geometry of the scene, and automatically produces high quality motions that achieve the desired user control. The entire network is trained in an end-to-end fashion on a large dataset composed of locomotion such as walking, running, jumping, and climbing movements fitted into virtual environments. Our system can therefore automatically produce motions where the character adapts to different geometric environments such as walking and running over rough terrain, climbing over large rocks, jumping over obstacles, and crouching under low ceilings. Our network architecture produces higher quality results than time-series autoregressive models such as LSTMs as it deals explicitly with the latent variable of motion relating to the phase. Once trained, our system is also extremely fast and compact, requiring only milliseconds of execution time and a few megabytes of memory, even when trained on gigabytes of motion data. Our work is most appropriate for controlling characters in interactive scenes such as computer games and virtual reality systems.



26:45

Phase-functioned neural networks for character control (SIGGRAPH 2017 Presentation)

3.5K views

3:32

4 Experiments Where the AI Outsmarted Its Creators 🤖

1M view

5:21

Flexible Muscle-Based Locomotion for Bipedal Creatures

1.6M views

7:02

SIGGRAPH 2018: DeepMimic paper (main video)

131K views

1:01:31

IK Rig: Procedural Pose Animation

118K views

1:51

Google's DeepMind AI Just Taught Itself To Walk

10M views

6:06

Locomotion Skills for Simulated Quadrupeds

212K views

8:49

🖥️ WRITING MY FIRST MACHINE LEARNING GAME! (1/4)

1.6M views

7:48

[SIGGRAPH Asia 2019] Neural State Machine for Character-Scene Interactions

58K views

10:01

Simulating Natural Selection

7.3M views

1:07:54

I3D'19 Keynote: "Machine Learning for Motion Synthesis and Character Control" (Michael Buttner)

5.2K views

5:58

MarI/O - Machine Learning for Video Games

9.5M views

6:07

[SIGGRAPH 2018] Mode-Adaptive Neural Networks for Quadruped Motion Control

130K views

6:37

UE4 - Advanced Locomotion System V3 - Features

192K views

3:29

Evolution of Neural Networks using Genetic Algorithm for a 3D car made in Unity

645K views

3:26

Google DeepMind AI Does Parkour - Producing flexible behaviours in simulated environments

239K views

4:21

A Deep Learning Framework for Character Motion Synthesis and Editing (SIGGRAPH 2016)

50K views