Watch Your Step: Real-Time Adaptive Character Stepping

TL;DR

This paper introduces a fast, robust, and easy-to-implement real-time control algorithm for bipedal character stepping, combining a modified inverted-pendulum model with ankle feedback and variable leg lengths to handle diverse interactive scenarios.

Contribution

The authors develop a novel extension of the inverted-pendulum model that enables real-time adaptive stepping for humanoid avatars without pre-processing or online optimization.

Findings

Handles terrain height changes and push forces effectively

Produces human-like, upright stepping motions

Operates in real-time with high computational efficiency

Abstract

An effective 3D stepping control algorithm that is computationally fast, robust, and easy to implement is extremely important and valuable to character animation research. In this paper, we present a novel technique for generating dynamic, interactive, and controllable biped stepping motions. Our approach uses a low-dimensional physics-based model to create balanced humanoid avatars that can handle a wide variety of interactive situations, such as terrain height shifting and push exertions, while remaining upright and balanced. We accomplish this by combining the popular inverted-pendulum model with an ankle-feedback torque and variable leg-length mechanism to create a controllable solution that can adapt to unforeseen circumstances in real-time without key-framed data, any offline pre-processing, or on-line optimizations joint torque computations. We explain and address oversimplifications and limitations with the basic IP model and the reasons for extending the model by means of additional control mechanisms. We demonstrate a simple and fast approach for extending the IP model based on an ankle-torque and variable leg lengths approximation without hindering the extremely attractive properties (i.e., computational speed, robustness, and simplicity) that make the IP model so ideal for generating upright responsive balancing biped movements. Finally, while our technique focuses on lower body motions, it can, nevertheless, handle both small and large push forces even during terrain height variations. Moreover, our model effectively creates human-like motions that synthesize low-level upright stepping movements, and can be combined with additional controller techniques to produce whole body autonomous agents.