Recurrent Transition Networks for Character Locomotion

TL;DR

This paper introduces Recurrent Transition Networks, a deep learning approach using LSTM-based models to automatically generate realistic character transition animations, improving efficiency and quality in complex locomotion systems.

Contribution

The paper presents a novel LSTM-based recurrent neural network architecture for automatic transition generation, trained without explicit gait or contact labels, and enhanced with terrain-awareness for better rough-terrain navigation.

Findings

Generated transitions rival motion capture quality

Terrain-aware input improves rough-terrain navigation

Model successfully reconstructs high-quality motions from sparse data

Abstract

Manually authoring transition animations for a complete locomotion system can be a tedious and time-consuming task, especially for large games that allow complex and constrained locomotion movements, where the number of transitions grows exponentially with the number of states. In this paper, we present a novel approach, based on deep recurrent neural networks, to automatically generate such transitions given a past context of a few frames and a target character state to reach. We present the Recurrent Transition Network (RTN), based on a modified version of the Long-Short-Term-Memory (LSTM) network, designed specifically for transition generation and trained without any gait, phase, contact or action labels. We further propose a simple yet principled way to initialize the hidden states of the LSTM layer for a given sequence which improves the performance and generalization to new motions. We both quantitatively and qualitatively evaluate our system and show that making the network terrain-aware by adding a local terrain representation to the input yields better performance for rough-terrain navigation on long transitions. Our system produces realistic and fluid transitions that rival the quality of Motion Capture-based ground-truth motions, even before applying any inverse-kinematics postprocess. Direct benefits of our approach could be to accelerate the creation of transition variations for large coverage, or even to entirely replace transition nodes in an animation graph. We further explore applications of this model in a animation super-resolution setting where we temporally decompress animations saved at 1 frame per second and show that the network is able to reconstruct motions that are hard to distinguish from un-compressed locomotion sequences.