Optimizing Locomotion Controllers Using Biologically-Based Actuators and Objectives

By Jack M. Wang, Samuel R. Hamner, Scott L. Delp, Vladlen Koltun. ACM Transactions on Graphics, 2012

313 citations recorded by the source corpus.

We present a technique for automatically synthesizing walking and running controllers for physically-simulated 3D humanoid characters. The sagittal hip, knee, and ankle degrees-of-freedom are actuated using a set of eight Hill-type musculotendon models in each leg, with biologically-motivated control laws. The parameters of these control laws are set by an optimization procedure that satisfies a number of locomotion task terms while minimizing a biological model of metabolic energy expenditure. We show that the use of biologically-based actuators and objectives measurably increases the realism of gaits generated by locomotion controllers that operate without the use of motion capture data, and that metabolic energy expenditure provides a simple and unifying measurement of effort that can be used for both walking and running control optimization.

Subfields: Computer Vision and Pattern Recognition, Control and Systems Engineering

View Paper: https://www.semanticscholar.org/paper/cca186653ef879b835861245cd97b356bc629cd9

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Control and Systems Engineering

Citations in the corpus

AMP (2021, background): “Simulating more biologically accurate actuators can also improve motion quality [Geijtenbeek et al. 2013; Jiang et al. 2019; Wang et al. 2012], but may nonetheless yield unnatural behaviors.”
Learning locomotion skills using DeepRL: does the choice of action space matter? (2016, methodology): “Biologically-inspired muscle models have also been explored for locomotion [Wang et al. 2012].”
Combining model-based policy search with online model learning for control of physical humanoids (2016, background): “Researchers in biomechanics and graphics have found that explicitly incorporating sensory delays has an impact on the resulting movement behaviors [6], [27]”
Is my model good enough? Best practices for verification and validation of musculoskeletal models and simulations of movement. (2015, background): “Further, recent work in predictive simulation has shown that muscle–tendon dynamics are required to achieve humanlike motion [89,90].”
DeepMimic (2018, background): “Many such controllers are the products of an underlying simplified model and an optimization process, where a compact set of parameters are tuned in order to achieve the desired behaviors [Agrawal et al. 2013; Ha and Liu 2014; Wang et al. 2012].”
Fluid directed rigid body control using deep reinforcement learning (2018, methodology): “Such techniques have been used to control low-DOF dynamic systems, including articulated characters [Wang et al. 2012; Yin et al. 2007].”
Learning symmetric and low-energy locomotion (2018, methodology): “One approach is the use of finite state machines (FSMs) [Coros et al. 2010, 2011; de Lasa et al. 2010; Felis and Mombaur 2016; Geijtenbeek et al. 2013; Hodgins et al. 1995; Jain et al. 2009; Wang et al. 2009, 2012; Yin et al. 2007].”