Nalin Harischandra

Sensory feedback plays a significant role in generating walking gait and in gait transition in salamanders : a simulation study

Here, we investigate the role of sensory feedback in gait generation and transition by using a three-dimensional, neuro-musculo-mechanical model of a salamander with realistic physical parameters. Activation of limb and axial muscles were driven by neural output patterns obtained from a central pattern generator (CPG) which is composed of simulated spiking neurons with adaptation. The CPG consists of a body-CPG and four limb-CPGs that are interconnected via synapses both ipsilaterally and contralaterally. We use the model both with and without sensory modulation and four different combinations of ipsilateral and contralateral coupling between the limb-CPGs. We found that the proprioceptive sensory inputs are essential in obtaining a coordinated lateral sequence walking gait (walking). The sensory feedback includes the signals coming from the stretch receptor like intraspinal neurons located in the girdle regions and the limb stretch receptors residing in the hip and scapula regions of the salamander. On the other hand, walking trot gait (trotting) is more under central (CPG) influence compared to that of the peripheral or sensory feedback. We found that the gait transition from walking to trotting can be induced by increased activity of the descending drive coming from the mesencephalic locomotor region and is helped by the sensory inputs at the hip and scapula regions detecting the late stance phase. More neurophysiological experiments are required to identify the precise type of mechanoreceptors in the salamander and the neural mechanisms mediating the sensory modulation.

Decoding the mechanisms of gait generation in salamanders by combining neurobiology, modeling and robotics

Vertebrate animals exhibit impressive locomotor skills. These locomotor skills are due to the complex interactions between the environment, the musculo-skeletal system and the central nervous system, in particular the spinal locomotor circuits. We are interested in decoding these interactions in the salamander, a key animal from an evolutionary point of view. It exhibits both swimming and stepping gaits and is faced with the problem of producing efficient propulsive forces using the same musculo-skeletal system in two environments with significant physical differences in density, viscosity and gravitational load. Yet its nervous system remains comparatively simple. Our approach is based on a combination of neurophysiological experiments, numerical modeling at different levels of abstraction, and robotic validation using an amphibious salamander-like robot. This article reviews the current state of our knowledge on salamander locomotion control, and presents how our approach has allowed us to obtain a first conceptual model of the salamander spinal locomotor networks. The model suggests that the salamander locomotor circuit can be seen as a lamprey-like circuit controlling axial movements of the trunk and tail, extended by specialized oscillatory centers controlling limb movements. The interplay between the two types of circuits determines the mode of locomotion under the influence of sensory feedback and descending drive, with stepping gaits at low drive, and swimming at high drive.

A 3D Musculo-Mechanical Model of the Salamander for the Study of Different Gaits and Modes of Locomotion.

Computer simulation has been used to investigate several aspects of locomotion in salamanders. Here we introduce a three-dimensional forward dynamics mechanical model of a salamander, with physically realistic weight and size parameters. Movements of the four limbs and of the trunk and tail are generated by sets of linearly modeled skeletal muscles. In this study, activation of these muscles were driven by prescribed neural output patterns. The model was successfully used to mimic locomotion on level ground and in water. We compare the walking gait where a wave of activity in the axial muscles travels between the girdles, with the trotting gait in simulations using the musculo-mechanical model. In a separate experiment, the model is used to compare different strategies for turning while stepping; either by bending the trunk or by using side-stepping in the front legs. We found that for turning, the use of side-stepping alone or in combination with trunk bending, was more effective than the use of trunk bending alone. We conclude that the musculo-mechanical model described here together with a proper neural controller is useful for neuro-physiological experiments in silico.

System identification of muscle–joint interactions of the cat hind limb during locomotion

Neurophysiological experiments in walking cats have shown that a number of neural control mechanisms are involved in regulating the movements of the hind legs during locomotion. It is experimentally hard to isolate individual mechanisms without disrupting the natural walking pattern and we therefore introduce a different approach where we use a model to identify what control is necessary to maintain stability in the musculo-skeletal system. We developed a computer simulation model of the cat hind legs in which the movements of each leg are produced by eight limb muscles whose activations follow a centrally generated pattern with no proprioceptive feedback. All linear transfer functions, from each muscle activation to each joint angle, were identified using the response of the joint angle to an impulse in the muscle activation at 65 postures of the leg covering the entire step cycle. We analyzed the sensitivity and stability of each muscle action on the joint angles by studying the gain and pole plots of these transfer functions. We found that the actions of most of the hindlimb muscles display inherent stability during stepping, even without the involvement of any proprioceptive feedback mechanisms, and that those musculo-skeletal systems are acting in a critically damped manner, enabling them to react quickly without unnecessary oscillations. We also found that during the late swing, the activity of the posterior biceps/semitendinosus (PB/ST) muscles causes the joints to be unstable. In addition, vastus lateralis (VL), tibialis anterior (TA) and sartorius (SAT) muscle–joint systems were found to be unstable during the late stance phase, and we conclude that those muscles require neuronal feedback to maintain stable stepping, especially during late swing and late stance phases. Moreover, we could see a clear distinction in the pole distribution (along the step cycle) for the systems related to the ankle joint from that of the other two joints, hip or knee. A similar pattern, i.e., a pattern in which the poles were scattered over the s-plane with no clear clustering according to the phase of the leg position, could be seen in the systems related to soleus (SOL) and TA muscles which would indicate that these muscles depend on neural control mechanisms, which may involve supraspinal structures, over the whole step cycle.

A computational model of visually guided locomotion in lamprey

This study addresses mechanisms for the generation and selection of visual behaviors in anamniotes. To demonstrate the function of these mechanisms, we have constructed an experimental platform where a simulated animal swims around in a virtual environment containing visually detectable objects. The simulated animal moves as a result of simulated mechanical forces between the water and its body. The undulations of the body are generated by contraction of simulated muscles attached to realistic body components. Muscles are driven by simulated motoneurons within networks of central pattern generators. Reticulospinal neurons, which drive the spinal pattern generators, are in turn driven directly and indirectly by visuomotor centers in the brainstem. The neural networks representing visuomotor centers receive sensory input from a simplified retina. The model also includes major components of the basal ganglia, as these are hypothesized to be key components in behavior selection. We have hypothesized that sensorimotor transformation in tectum and pretectum transforms the place-coded retinal information into rate-coded turning commands in the reticulospinal neurons via a recruitment network mimicking the layered structure of tectal areas. Via engagement of the basal ganglia, the system proves to be capable of selecting among several possible responses, even if exposed to conflicting stimuli. The anatomically based structure of the control system makes it possible to disconnect different neural components, yielding concrete predictions of how animals with corresponding lesions would behave. The model confirms that the neural networks identified in the lamprey are capable of responding appropriately to simple, multiple, and conflicting stimuli.

Insect-inspired tactile contour sampling using vibration-based robotic antennae

Compared to vision, active tactile sensing enables animals and robots to perform unambiguous object localization, segmentation and shape recognition. Recently, we proposed a bio-inspired, CPG-based, active antennal control model, so-called Contour-net, which captures essential characteristics of antennal behavior in climbing stick insects. In simulation, this model provides a robust and effective way to trace contours and classify various 3D shapes. Here, we propose a physical robotic implementation of Contour-net using vibration-based active antennae. We show that combining tactile contour tracing with vibration-based distance estimation yields fairly accurate localization of contact events in 3D space.

Stable phase-shift despite quasi-rhythmic movements: a CPG-driven dynamic model of active tactile exploration in an insect.

An essential component of autonomous and flexible behavior in animals is active exploration of the environment, allowing for perception-guided planning and control of actions. An important sensory system involved is active touch. Here, we introduce a general modeling framework of Central Pattern Generators (CPGs) for movement generation in active tactile exploration behavior. The CPG consists of two network levels: (i) phase-coupled Hopf oscillators for rhythm generation, and (ii) pattern formation networks for capturing the frequency and phase characteristics of individual joint oscillations. The model captured the natural, quasi-rhythmic joint kinematics as observed in coordinated antennal movements of walking stick insects. Moreover, it successfully produced tactile exploration behavior on a three-dimensional skeletal model of the insect antennal system with physically realistic parameters. The effect of proprioceptor ablations could be simulated by changing the amplitude and offset parameters of the joint oscillators, only. As in the animal, the movement of both antennal joints was coupled with a stable phase difference, despite the quasi-rhythmicity of the joint angle time courses. We found that the phase-lead of the distal scape-pedicel (SP) joint relative to the proximal head-scape (HS) joint was essential for producing the natural tactile exploration behavior and, thus, for tactile efficiency. For realistic movement patterns, the phase-lead could vary within a limited range of 10–30° only. Tests with artificial movement patterns strongly suggest that this phase sensitivity is not a matter of the frequency composition of the natural movement pattern. Based on our modeling results, we propose that a constant phase difference is coded into the CPG of the antennal motor system and that proprioceptors are acting locally to regulate the joint movement amplitude.

Shape Recognition Through Tactile Contour Tracing

We present Contour-net, a bio-inspired model for tactile contour-tracing driven by an Hopf oscillator. By controlling the rhythmic movements of a simulated insect-like feeler, the model executes both wide searching and local sampling movements. Contour-tracing is achieved by means of contact-induced phase-forwarding of the oscillator. To classify the shape of an object, collected contact events can be directly fed into machine learning algorithms with minimal pre-processing (scaling). Three types of classifiers were evaluated, the best one being a Support Vector Machine. The likelihood of correct classification steadily increases with the number of collected contacts, enabling an incremental classification during sampling. Given a sufficiently large training data set, tactile shape recognition can be achieved in a position-, orientation- and size-invariant manner. The suitability for robotic applications is discussed.

A forward model for an active tactile sensor using Echo State Networks

Here, we introduce a forward model designed for predicting the expected reading of a bionic tactile sensor (antenna) mounted onto a wheeled robot. The model was used to distinguish self-generated stimulation from true tactile events to the antenna. An Echo State Network (ESN), a special type of recurrent neural network which is suitable for chaotic time series prediction, is used to implement the forward model. Inputs to the ESN are the motor command which sets the position of the antenna, and a local proprioceptive signal which measures the acceleration of the robot platform. The model can successfully be used to detect a tactile contact on the antenna while the robot is moving along a path with obstacles. Such forward models are good candidates to be used in neural yet simple way to eliminate self-stimulation of sensors of other modalities due to ego-motion.