Generative GaitNet

TL;DR

Generative GaitNet is a deep reinforcement learning-based system that models the complex relationship between anatomy and gait, capable of generating diverse healthy and pathological gait patterns in real-time physics simulations.

Contribution

It introduces a novel neural network architecture that integrates anatomy and gait conditions to produce realistic gait cycles through physics-based simulation.

Findings

Generates diverse gait patterns in real-time

Models complex anatomy-gait relationships

Demonstrates efficacy in physics-based simulation

Abstract

Understanding the relation between anatomy andgait is key to successful predictive gait simulation. Inthis paper, we present Generative GaitNet, which isa novel network architecture based on deep reinforce-ment learning for controlling a comprehensive, full-body, musculoskeletal model with 304 Hill-type mus-culotendons. The Generative Gait is a pre-trained, in-tegrated system of artificial neural networks learnedin a 618-dimensional continuous domain of anatomyconditions (e.g., mass distribution, body proportion,bone deformity, and muscle deficits) and gait condi-tions (e.g., stride and cadence). The pre-trained Gait-Net takes anatomy and gait conditions as input andgenerates a series of gait cycles appropriate to theconditions through physics-based simulation. We willdemonstrate the efficacy and expressive power of Gen-erative GaitNet to generate a variety of healthy andpathologic human gaits in real-time physics-based sim-ulation.