Biped gait controller for large speed variations, combining reflexes and a central pattern generator in a neuromuscular model

Noot, Nicolas Van der; Ijspeert, Auke Jan; Ronsse, Renaud

doi:10.1109/icra.2015.7140079

Cited by 49 publications

(32 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…where K swing p is the position gain on the leg angle, K swing d is the velocity gain, α is the leg angle andα is the leg angular velocity (see Figure 3). More concisely, the swing parameters that we focus on in our optimization are the following: 1) K swing p : Position gain on feedback on leg angle 2) K swing d : Velocity gain on feedback on leg velocity 3) α 0 : Nominal leg angle 4) C d : Gain on the horizontal distance between the stance foot and CoM 5) C v : Gain on the horizontal velocity of the CoM 6) l clr : Desired leg clearance Though originally developed for explaining human neural control pathways, these controllers have recently been applied to robots and prosthetics, for example in [19] and [20]. As demonstrated in [18], these models are indeed capable of generating a variety of locomotion behaviours for a humanoid model -for example, walking on flat, rough ground, turning, running, walking upstairs and on ramps.…”

Section: B Optimization For Bipedal Locomotionmentioning

confidence: 99%

Sample efficient optimization for learning controllers for bipedal locomotion

Antonova

Rai

Atkeson

2016

2016 IEEE-RAS 16th International Conference on Humanoid Robots (Humanoids)

View full text Add to dashboard Cite

Abstract-Learning policies for bipedal locomotion can be difficult, as experiments are expensive and simulation does not usually transfer well to hardware. To counter this, we need algorithms that are sample efficient and inherently safe. Bayesian Optimization is a powerful sample-efficient tool for optimizing non-convex black-box functions. However, its performance can degrade in higher dimensions. We develop a distance metric for bipedal locomotion that enhances the sample-efficiency of Bayesian Optimization and use it to train a 16 dimensional neuromuscular model for planar walking. This distance metric reflects some basic gait features of healthy walking and helps us quickly eliminate a majority of unstable controllers. With our approach we can learn policies for walking in less than 100 trials for a range of challenging settings. In simulation, we show results on two different costs and on various terrains including rough ground and ramps, sloping upwards and downwards. We also perturb our models with unknown inertial disturbances analogous with differences between simulation and hardware. These results are promising, as they indicate that this method can potentially be used to learn control policies on hardware.

show abstract

Section: B Optimization For Bipedal Locomotionmentioning

confidence: 99%

Sample efficient optimization for learning controllers for bipedal locomotion

Antonova

Rai

Atkeson

2016

2016 IEEE-RAS 16th International Conference on Humanoid Robots (Humanoids)

View full text Add to dashboard Cite

show abstract

“…Common muscle synergies exist between walking and standing (Chvatal and Ting, 2013 ), meaning that one neural control model could potentially be applied to walking and standing. Previously, a reflex controller was shown to replicate normal human walking (Geyer and Herr, 2010 ; Song and Geyer, 2015 ), which has since been applied to control exoskeletons (Wu et al, 2017 ), prostheses (Eilenberg et al, 2010 ), and a humanoid robot (Van der Noot et al, 2015 ), and was extended, for example to include frontal muscles as well (Song and Geyer, 2015 ).…”

Section: Introductionmentioning

confidence: 99%

Exploring the Contribution of Proprioceptive Reflexes to Balance Control in Perturbed Standing

Koelewijn

Ijspeert

2020

Front. Bioeng. Biotechnol.

Self Cite

View full text Add to dashboard Cite

Humans control balance using different feedback loops involving the vestibular system, the visual system, and proprioception. In this article, we focus on proprioception and explore the contribution of reflexes based on force and length feedback to standing balance. In particular, we address the questions of how much proprioception alone could explain balance control, and whether one modality, force or length feedback, is more important than the other. A sagittal plane neuro-musculoskeletal model was developed with six degrees of freedom and nine muscles in each leg. A controller was designed using proprioceptive reflexes and a dead zone. No feedback control was applied inside the dead zone. Reflexes were active once the center of mass moved outside the dead zone. Controller parameters were found by solving an optimization problem, where effort was minimized while the neuro-musculoskeletal model should remain standing upright on a perturbed platform. The ground was perturbed with random square pulses in the sagittal plane with different amplitudes and durations. The optimization was solved for three controllers: using force and length feedback (base model), using only force feedback, and using only length feedback. Simulations were compared to human data from previous work, where an experiment with the same perturbation signal was performed. The optimized controller yielded a similar posture, since average joint angles were within 5 degrees of the experimental average joint angles. The joint angles of the base model, the length only model, and the force only model correlated weakly (ankle) to moderately with the experimental joint angles. The ankle moment correlated weakly to moderately with the experimental ankle moment, while the hip and knee moment were only weakly correlated, or not at all. The time series of the joint angles showed that the length feedback model was better able to explain the experimental joint angles than the force feedback model. Changes in time delay affected the correlation of the joint angles and joint moments. The objective of effort minimization yielded lower joint moments than in the experiment, suggesting that other objectives are also important in balance control, which cause an increase in effort and thus larger joint moments.

show abstract

“…Some models considered the CPGs as a generator of rhythmic motor commands [29, 30]. Some of CPG-based models represent hierarchical controller inspired by the biological structure of human motion [31], especially by coupling reflexes to the CPGs to control biped motion [32, 33]. One of the effects of reflexes on CPGs is known as phase resetting [34].…”

Section: Introductionmentioning

confidence: 99%

A simple CPG-based model to generate human hip moment pattern in walking by generating stiffness and equilibrium point trajectories

Bahramian

Towhidkhah

Jafari

2019

Preprint

View full text Add to dashboard Cite

Equilibrium point hypothesis (its developed version named as referent control theory)presents a theory about how the central nerves system (CNS) generates human movements. On the other hand, it has been shown that nerves circuits known as central pattern generators (CPG) likely produce motor commands to the muscles in rhythmic motions. In the present study, we designed a bio-inspired walking model, by coupling double pendulum to CPGs that produces equilibrium and stiffness trajectories as reciprocal and co-activation commands. As a basic model, it is has been shown that this model can regenerate pattern of a hip moment in the swing phase by high correlation ሺ ߩ ൌ 0 . 9 7 0 ሻ with experimental data. Moreover, it has been reported that a global electromyography (EMG) minima occurs in the mid-swing phase when the hip is more flexed in comparison with the other leg. Our model showed that equilibrium and actual hip angle trajectories match each other in mid-swing, similar to the mentioned posture, that is consistent with previous findings. Such a model can be used in active exoskeletons and prosthesis to make proper active stiffness and torque.controls the body?" [3, 4]. These theories explain how the CNS can release from complex calculations to control movements. According to EPH, at the muscle level, the CNS only determines a specific threshold (ߣ) for muscle [5, 6]. If actual muscle length goes beyond the defined threshold, EMG activity emerges as a result of the gap between actual and threshold muscle lengths. To control the joint, two commands are defined by the CNS: reciprocal (R) command and co-activation (C) command [7]. If joint is considered as rotational spring, R and C commands determine equilibrium point and stiffness of the joint, respectively. In accordance with the referent control theory, for multi-joint system (e.g. leg), it is assumed that initially, global R and C commands are set by CNS, then, by few-to-many mappings, r and c commands for various joints are defined. Finally, ߣ s for agonist and antagonist muscles emerged from these r and c by another mapping [8]. This hierarchical system is effective for making coordination between different joints and muscles [9]. One of the powerful experimental evidence for this hierarchical referent system is the global EMG minima [10]. When the referent (R) and actual configurations approach meet each other, a minimum EMG activity occurs for most of the muscles involved in the task [10]. This phenomena has been confirmed by many studies on various tasks such as walking and jumping [10], monkeys head movement [11], Jete in ballet dancers [12], and hammering [13]. It is shown that during stepping forward, a global EMG minima occurred during mid-swing when the hip flexion of the swing leg is slightly more than the other leg [14].On the other hand, the behavior of the joints as a rotational spring have commonly been used for investigating the relationship between torque(s) and angle(s). These studies are applicable to design anthropomorphic biped robots [15][16][...

show abstract

Biped gait controller for large speed variations, combining reflexes and a central pattern generator in a neuromuscular model

Cited by 49 publications

References 23 publications

Sample efficient optimization for learning controllers for bipedal locomotion

Sample efficient optimization for learning controllers for bipedal locomotion

Exploring the Contribution of Proprioceptive Reflexes to Balance Control in Perturbed Standing

A simple CPG-based model to generate human hip moment pattern in walking by generating stiffness and equilibrium point trajectories

Contact Info

Product

Resources

About