Search for: All records

During walking, sensory information is measured and monitored by sensory organs that can be found on and within various limb segments. Strain can be monitored by insect load sensors, campaniform sensilla (CS), which have components embedded within the exoskeleton. CS vary in eccentricity, size, and orientation, which can affect their sensitivity to specific strains. Directly investigating the mechanical interfaces that these sensors utilize to encode changes in load bears various obstacles, such as modeling of viscoelastic properties. To circumvent the difficulties of modeling and performing biological experiments in small insects, we developed 3-dimensional printed resin models based on high-resolution imaging of CS. Through the utilization of strain gauges and a motorized tensile tester, physiologically plausible strain can be mimicked while investigating the compression and tension forces that CS experience; here, this was performed for a field of femoral CS inDrosophila melanogaster. Different loading scenarios differentially affected CS compression and the likely neuronal activity of these sensors and elucidate population coding of stresses acting on the cuticle.

 

Discharges of sensory receptors (campaniform sensilla) in the hind legs of stick insects can differentially signal forces that occur in walking uphill versus walking downhill. Unexpectedly, sensory firing most closely reflects the rate of change of force (dF/d t) even when the force levels are high. These signals have been replicated in a mathematical model of the receptors and could be used to stabilize leg movements both in the animal and in a walking robot. 

A challenge in robotics is to control interactions with the environment by modulating the stiffness of a manipulator’s joints. Smart servos are controlled with proportional feedback gain that is analogous to torsional stiffness of an animal’s joint. In animals, antagonistic muscle groups can be temporarily coactivated to stiffen the joint to provide greater opposition to external forces. However, the joint properties for which coactivation increases the stiffness of the joint remain unknown. In this study, we explore possible mechanisms by building a mathematical model of the stick insect tibia actuated by two muscles, the extensor and flexor tibiae. Muscle geometry, passive properties, and active properties are extracted from the literature. Joint stiffness is calculated by tonically activating the antagonists, perturbing the joint from its equilibrium angle, and calculating the restoring moment generated by the muscles. No reflexes are modeled. We estimate how joint stiffness depends on parallel elastic element stiffness, the shape of the muscle activation curve, and properties of the force-length curve. We find that co-contracting antagonist muscles only stiffens the joint when the peak of the force-length curve occurs at a muscle length longer than that when the joint is at equilibrium and the muscle force versus activation curve is concave-up. We discuss how this information could be applied to the control of a smart servo actuator in a robot leg. 

A challenge in robotics is to control interactions with the environment by modulating the stiffness of a manipulator’s joints. Smart servos are controlled with proportional feedback gain that is analogous to torsional stiffness of an animal’s joint. In animals, antagonistic muscle groups can be temporarily coactivated to stiffen the joint to provide greater opposition to external forces. However, the joint properties for which coactivation increases the stiffness of the joint remain unknown. In this study, we explore possible mechanisms by building a mathematical model of the stick insect tibia actuated by two muscles, the extensor and flexor tibiae. Muscle geometry, passive properties, and active properties are extracted from the literature. Joint stiffness is calculated by tonically activating the antagonists, perturbing the joint from its equilibrium angle, and calculating the restoring moment generated by the muscles. No reflexes are modeled. We estimate how joint stiffness depends on parallel elastic element stiffness, the shape of the muscle activation curve, and properties of the force-length curve. We find that co-contracting antagonist muscles only stiffens the joint when the peak of the force-length curve occurs at a muscle length longer than that when the joint is at equilibrium and the muscle force versus activation curve is concave-up. We discuss how this information could be applied to the control of a smart servo actuator in a robot leg. 

Insects use various sensory organs to monitor proprioceptive and exteroceptive information during walking. The measurement of forces in the exoskeleton is facilitated by campaniform sensilla (CS), which monitor resisted muscle forces through the detection of exoskeletal strains. CS are commonly found in leg segments arranged in fields, groups, or as single units. Most insects have the highest density of sensor locations on the trochanter, a proximal leg segment. CS are arranged homologously across species, suggesting comparable functions despite noted morphological differences. Furthermore, the trochanter–femur joint is mobile in some species and fused in others. To investigate how different morphological arrangements influence strain sensing in different insect species, we utilized two robotic models of the legs of the fruit fly Drosophila melanogaster and the stick insect Carausius morosus. Both insect species are past and present model organisms for unraveling aspects of motor control, thus providing extensive information on sensor morphology and, in-part, function. The robotic models were dynamically scaled to the legs of the insects, with strain gauges placed with correct orientations according to published data. Strains were detected during stepping on a treadmill, and the sensor locations and leg morphology played noticeable roles in the strains that were measured. Moreover, the sensor locations that were absent in one species relative to the other measured strains that were also being measured by the existing sensors. These findings contributed to our understanding of load sensing in animal locomotion and the relevance of sensory organ morphology in motor control. 

One developing approach for robotic control is the use of networks of dynamic neurons connected with conductance-based synapses, also known as Synthetic Nervous Systems (SNS). These networks are often developed using cyclic topologies and heterogeneous mixtures of spiking and non-spiking neurons, which is a difficult proposition for existing neural simulation software. Most solutions apply to either one of two extremes, the detailed multi-compartment neural models in small networks, and the large-scale networks of greatly simplified neural models. In this work, we present our open-source Python package SNS-Toolbox, which is capable of simulating hundreds to thousands of spiking and non-spiking neurons in real-time or faster on consumer-grade computer hardware. We describe the neural and synaptic models supported by SNS-Toolbox, and provide performance on multiple software and hardware backends, including GPUs and embedded computing platforms. We also showcase two examples using the software, one for controlling a simulated limb with muscles in the physics simulator Mujoco, and another for a mobile robot using ROS. We hope that the availability of this software will reduce the barrier to entry when designing SNS networks, and will increase the prevalence of SNS networks in the field of robotic control. 

Sensory feedback from sense organs during animal locomotion can be heavily influenced by an organism’s mechanical structure. In insects, the interplay between sensing and mechanics can be demonstrated in the campaniform sensilla (CS) strain sensors located across the exoskeleton. Leg CS are highly sensitive to the loading state of the limb. In walking, loading is primarily influenced by ground reaction forces (GRF) mediated by the foot, or tarsus. The complex morphology of the tarsus provides compliance, passive and active substrate grip, and an increased moment arm for the GRF, all of which impact leg loading and the resulting CS discharge. To increase the biomimicry of robots we use to study strain feedback during insect walking, we have developed a series of tarsi for our robotic model of a Carausius morosus middle leg. We seek the simplest design that mimics tarsus functionality. Tarsi were designed with varying degrees of compliance, passive grip, and biomimetic structure. We created elastic silicone tarsal joints for several of these models and found that they produced linear stiffness within joint limits across different joint morphologies. Strain gauges positioned in CS locations on the trochanterofemur and tibia recorded strain while the leg stepped on a treadmill. Most, but not all, designs increased axial strain magnitude compared to previous data with no tarsus. Every tarsus design produced positive transversal strain in the tibia, indicating axial torsion in addition to bending. Sudden increases in tibial strain reflected leg slipping during stance. This data show how different aspects of the tarsus may mediate leg loading, allowing us to improve the mechanical biomimicry of future robotic test platforms. 

Animals utilize a number of neuronal systems to produce locomotion. One type of sensory organ that contributes in insects is the campaniform sensillum (CS) that measures the load on their legs. Groups of the receptors are found on high stress regions of the leg exoskeleton and they have significant effects in adapting walking behavior. Recording from these sensors in freely moving animals is limited by technical constraints. To better understand the load feedback signaled by CS to the nervous system, we have constructed a dynamically scaled robotic model of the Carausius morosus stick insect middle leg. The leg steps on a treadmill and supports weight during stance to simulate body weight. Strain gauges were mounted in the same positions and orientations as four key CS groups (Groups 3, 4, 6B, and 6A). Continuous data from the strain gauges were processed through a previously published dynamic computational model of CS discharge. Our experiments suggest that under different stepping conditions (e.g., changing “body” weight, phasic load stimuli, slipping foot), the CS sensory discharge robustly signals increases in force, such as at the beginning of stance, and decreases in force, such as at the end of stance or when the foot slips. Such signals would be crucial for an insect or robot to maintain intra- and inter-leg coordination while walking over extreme terrain. 

It is unknown precisely how the nervous system of invertebrates combines multiple sensory inputs to calculate more abstract quantities, e.g., combining the angle of multiple leg joints to calculate the position of the foot relative to the body. In this paper, we suggest that non-spiking interneurons (NSIs) in the nervous system could calculate such quantities and construct a neuromechanical model to support the claim. Range fractionated sensory inputs are modeled as multiple integrate-and-fire neurons. The NSI is modeled as a multi-compartment dendritic tree and one large somatic compartment. Each dendritic compartment receives synaptic input from one sensory neuron from the knee and one from the hip. Every dendritic compartment connects to the soma. The model is constructed within the Animatlab 2 software. The neural representation of the system accurately follows the true position of the foot. We also discuss motivation for future research, which includes modeling other hypothetical networks in the insect nervous system and integrating this model into task-level robot control. 

Load sensing is critical for walking behavior in animals, who have evolved a number of sensory organs and neural systems to improve their agility. In particular, insects measure load on their legs using campaniform sensilla (CS), sensory neurons in the cuticle of high-stress portions of the leg. Extracellular recordings from these sensors in a behaving animal are difficult to collect due to interference from muscle potentials, and some CS groups are largely inaccessible due to their placement on the leg. To better understand what loads the insect leg experiences and what sensory feedback the nervous system may receive during walking, we constructed a dynamically-scaled robotic model of the leg of the stick insect Carausius morosus. We affixed strain gauges in the same positions and orientations as the major CS groups on the leg, i.e., 3, 4, 6A, and 6B. The robotic leg was mounted to a vertically-sliding linear guide and stepped on a treadmill to simulate walking. Data from the strain gauges was run through a dynamic model of CS discharge developed in a previous study. Our experiments reveal stereotypical loading patterns experienced by the leg, even as its weight and joint stiffness is altered. Furthermore, our simulated CS strongly signal the beginning and end of stance phase, two key events in the coordination of walking.