Bulletin of the American Physical Society
APS March Meeting 2021
Volume 66, Number 1
Monday–Friday, March 15–19, 2021; Virtual; Time Zone: Central Daylight Time, USA
Session R14: Robophysics: Robotics Meets Physics III: Limbless & Collective LocomotionFocus Live
|
Hide Abstracts |
Sponsoring Units: DBIO DSOFT Chair: PERRIN SCHIEBEL, Harvard University |
Thursday, March 18, 2021 8:00AM - 8:36AM Live |
R14.00001: Efficient sliding locomotion for two- and three-link bodies Invited Speaker: Silas Alben We study the possibility of efficient intermittent (inertia-based) locomotion for two-link bodies that open and close periodically in time. The anisotropy ratio of the sliding friction coefficients is a key parameter. With very anisotropic friction, efficient motions involve coasting in low-drag states, with rapid and asymmetric power and recovery strokes. As the anisotropy decreases, burst-and-coast motions change to motions with long power strokes and short recovery strokes. We find these motions in the spaces of sinusoidal and power-law motions described by two and five parameters, respectively, and with an optimization search in the space of more general periodic functions (truncated Fourier series). When we increase the resistive force's power-law dependence on velocity, the optimal motions become smoother, slower, and less efficient, particularly near isotropic friction. We then discuss efficient locomotion for three link bodies. |
Thursday, March 18, 2021 8:36AM - 8:48AM Live |
R14.00002: Terrestrial limbless gait selection through friction modulation Xiaotian Zhang, Noel Naughton, Tejaswin Parthasarathy, Mattia Gazzola Limbless locomotion is exhibited by a wide taxonomy of animals and has been observed in water, land, and even air. While broad principles underlying aquatic limbless locomotion have been articulated, a similar understanding for the terrestrial variety remain elusive. Here we show through a model snake how a variety of gaits resulting from both active 3D body deformations and interactions with heterogenous environments, can be all viewed as modulation of local frictional effects along the body. This simple perspective provides new understanding of how snakes and other limbless creatures may exploit their compliant body and naturally occurring substrate features to achieve adaptivity, control and gait selection. |
Thursday, March 18, 2021 8:48AM - 9:00AM Live |
R14.00003: Nematode omega turns improve reorientation in a limbless robot Tianyu Wang, Baxi Chong, Kelimar Diaz, Julian Whitman, Hang Lu, Daniel I Goldman, Howie Choset Limbless robots have the potential to locomote through confined spaces but turning effectively within unmodelled and unsensed environments remains challenging. Nematode worms like C. elegans use a turning strategy dubbed the "omega turn"; we posited that such turns would yield benefits in limbless robots. Study of the dynamics of the worm's turns reveals a sequence of entry, reversal, and exit phases and a superposition of two traveling waves. We use geometric mechanics tools to model the superposition of two traveling waves to generate effective body rotation while avoiding self-collision, and generate a scheme to reorient the robot to arbitrary angles via modulation of joint angle amplitudes. We experimentally test the omega turn on a one-meter-long limbless 16-segment robot on hard ground, validating that the omega turn gait outperforms previous turning gaits: it results in a larger angular displacement and a smaller area swept by the body over a gait cycle, allowing the robot to turn in highly confined spaces. Through robot experiment, we demonstrate that the joint angle amplitude modulation scheme allows the robot to reorient to arbitrary angles between 35 and 98 degrees in one gait cycle. |
Thursday, March 18, 2021 9:00AM - 9:12AM Live |
R14.00004: A new robophysical model for limbless locomotion reveals the importance of passive dynamics in obstacle navigation. Marine Maisonneuve, PERRIN SCHIEBEL, Kelimar Diaz, Daniel I Goldman Limbless animals like snakes and nematode worms move in complex terrain via propagation of body bending waves generated by alternating unilateral muscle activation. A previously studied desert specialist snake relies on passive body buckling, facilitated by unilateral activation, to traverse sparse obstacles. To discover principles by which passive body buckling can simplify control in complex terrain, we developed a robot (8 joints, 45 cm long) that models biological muscle morphology and head-to-tail propagating wave activation patterns. Each joint generates a bend using a pair of string actuators that shortens wires when active and allows lengthening when inactive. We studied the dynamics and locomotion of the model in both wall-collision and hexagonal lattice transit experiments. During wall collisions, the robot passively reoriented within a few undulation cycles and then moved parallel to the wall. In the lattice experiments, the robot often traversed the lattice without becoming jammed; these dynamics were generated by passive buckling and a spontaneous reversal behavior. The locomotor success of the robophysical model without need for sensory feedback suggests the importance of passive mechanisms in limbless organisms and can enhance robot locomotion in complex terrain. |
Thursday, March 18, 2021 9:12AM - 9:24AM Live |
R14.00005: Reconstruction of Backbone Curves for 3-D Locomotion of Limbless Robots Tianyu Wang, Bo Lin, Baxi Chong, Julian Whitman, Matthew Travers, Daniel I Goldman, Howie Choset, Greg Blekherman Limbless robots composed of alternating single-axis pitch and yaw joints have many internal degrees of freedom, which make them capable of versatile three-dimensional locomotion. Typically, motions are planned kinematically by a chronological and continuous sequence of backbone curves that capture desired macroscopic shapes of the robot. However, as the geometric arrangement of single-axis rotary joints creates constraints on the rotations in the robot, the robot typically does not accurately achieve the desired shapes defined by these backbone curves. This leads to locomotor dynamics which are not well predicted by geometric models due to undesired contact with the environment. Here we propose a method for snake robots to reconstruct desired backbone curves by solving a polynomial optimization problem that exploits the robot's geometric structure. The method enables accurate curve-configuration conversions for commonly used 3-D gaits like sidewinding, sinus lifting, and helical rolling. We also demonstrated via robot experiments that the method results in smooth locomotion of the robot - in a full gait cycle, we found that the existing method resulted in an average joint angle change between time steps of 10.42 degrees, whereas the ones for our method were only 1.64 degrees. |
Thursday, March 18, 2021 9:24AM - 9:36AM Live |
R14.00006: A sensorized robot to study physical interaction in limbless locomotion in complex 3-D terrain Divya Ramesh, Qiyuan Fu, Kaiwen Wang, Ratan Sadanand Othayoth Mullankandy, Chen Li Snakes can transition between different strategies quickly and stably using distributed body-terrain contact forces. Although snake robots hold the promise of doing the same, they are far from being able to move well in complex 3-D terrain with many large obstacles. We aim to understand how snakes sense and control physical interaction with the obstacles to generate effective locomotion. Towards this goal, we developed a sensorized snake robot physical model that uses a low-cost, flexible, piezoresistive sensor array distributed over its body to measure normal contact forces. Despite disturbance caused by self-deformation during the robot motion, calibration showed that the sensor array can measure forces of up to 10% body weight at 30 locations along the body at 10 Hz. We will study the robot as a physical model to understand the principles of how to control body deformation and terrain contact to generate forces while maintaining stability. This will not be informed by and in turn help understand our animal observation (see talk by Fu et al. How snakes traverse large obstacles in complex 3-D terrain). We are also exploring ways to increase the sensitivity of the sensor array and developing a sensorized complex terrain platform to measure distributed contact forces in snakes. |
Thursday, March 18, 2021 9:36AM - 9:48AM Live |
R14.00007: Nonlocal control of active agents on a deformable substrate Hussain Gynai, Shengkai Li, Yasemin Ozkan-Aydin, Steven Tarr, Enes Aydin, Pablo Laguna, Daniel I Goldman In an effort to construct a robophysical analog gravity system, we previously studied the dynamics of a 200-gram differential wheeled vehicle driving on a deformable spandex membrane (d=2.4m) with a static central depression (Li et al., 2019). The system displays rich dynamics, including precessing orbits reminiscent of those in general relativity. To take the next step and study how the vehicle dynamics can be manipulated via changes in membrane curvature, we developed an automated gantry system that can control the position and depth of a spherical object. The object creates a deformation on the membrane which influences the dynamics of the vehicle. Experiments reveal that forcing the object in circular orbits on the membrane can lead to “capture” of the moving vehicle depending on the initial orientation and position of the vehicle. Motivated by this, we augmented the vehicle with sensors which measure its instantaneous roll and pitch (tilt). A scheme which decreases the speed with tilt generally enhances capture, whereas, increasing speed with tilt generally prevents capture. In both cases, the trajectories of the vehicle are highly variable with even small changes to initial condition. A numerical model based on Poisson equation rationalizes our experimental observations. |
Thursday, March 18, 2021 9:48AM - 10:00AM Live |
R14.00008: Optimizing the Locomotion of a Robotic Active Matter System of Smarticles Annalisa Tulle Taylor, Thomas Alejandro Berrueta, Todd D. Murphey We consider the locomotive properties of an active matter system of shape-changing robots. This system is composed of smart, active particles, termed smarticles, that interact with each other via inter-robot collisions. In contrast to meticulously designed traditional robots, the smarticle robot swarm consists of generic, reconfigurable components noisily interacting. Although control of such a collective is beyond existing methods, the resulting system would be scalable, robust, and flexible in structure. Rather than adapting the model-agnostic formalisms of control theory, we posit that by considering the fundamental physics of the system, the swarm may become controllable. We have shown that when enclosed in a rigid ring, the smarticles exhibit emergent locomotion if the symmetry of internal configurations is broken by an inactive robot. Here, we use geometric mechanics to optimize robot inactivations to allow for faster directed locomotion in the plane. We demonstrate this in an experimental system of five smarticles and find that the system locomotes optimally when two smarticles are inactive. Finally, we find an expression for the optimal number of inactive smarticles as a function of the size of the swarm. |
Thursday, March 18, 2021 10:00AM - 10:12AM Live |
R14.00009: Synchronized swimming: collisions drive gait compatibility in undulatory robots Wei Zhou, Jaquelin Dezha-Peralta, Zhuonan Hao, Nick Gravish Many groups of organisms live in close proximity and are capable of complex collective movement. The locomotion of individual organisms, and thus bio-inspired robots, within the collective occur through periodic oscillation of internal degrees of freedom. It is a fundamental goal of swarm robotics, and more broadly active matter systems, to understand how effective collective behaviors can emerge from simple principles of interaction. In this presentation we demonstrate how groups of simple bio-inspired robots that move through lateral body undulation and interact only through contact forces can collectively locomote in close proximity. We introduce the idea of a compatible gait configuration, a metric that measures the ability for pairs of robots with fixed wavelength (λ) and phase difference (ΔΦ) to move in close proximity without colliding. Through experiment and simulation, we demonstrate that gait compatibility can be achieved through spatial re-configuration along the traveling wave direction, Δx=(λ/2π)ΔΦ. Critically, the ability to achieve compatibility is driven through contact interactions between adjacent robots within the gait cycle no long range interactions or communication is required. |
Thursday, March 18, 2021 10:12AM - 10:24AM Not Participating |
R14.00010: Emergent cooperation in ant-like robots Fabio Giardina, S Ganga Prasath, Souvik Mandal, Venkatesh Murthy, L. Mahadevan Carpenter ants are known to exhibit cooperative tunnel digging and bridge building. They rely on touch and chemical signaling (pheromones) for communication, but it is not known how these environmental cues are processed to drive cooperation. To understand the minimal number of sufficient behavioral rules for cooperation, we developed a robotic platform consisting of ant-like robots that can both physically interact and communicate through a virtual pheromone field. A mechanism enables the robot ants to manipulate the environment to either create a bridge or construct a tunnel. We present simple behavioral rules that lead to emerging cooperation in our robotic platform and will show how these insights can inform us on the internal processes that drive collective construction in biological systems. |
Thursday, March 18, 2021 10:24AM - 10:36AM Live |
R14.00011: When the mob sees the light - distributed learning of phototaxis without local gradients using robot morphology Matan Yah Ben Zion, Yoones Mirhosseini, Nicolas Bredeche, Olivier Dauchot A typical phototactic strategy requires a local intensity gradient monitoring, either through fore-aft differential light sensing, or through temporal differentiation on the fly. Using their sheer size, schools and flocks can respond to minute gradients the small individual could not detect accurately on its own. For this, the swarm needs cohesivity (through an effective attraction or an alignment interaction) in order to maintain a continuously connected communication network. We show that a completely decentralized population of kilobots can collectively find a successful phototactic strategy even in the total absence of a local gradient-field. Using an unsupervised online learning algorithm, the robots can reach a consensus for a phototactic policy, while their communication network remains sparse, intermittent, and disconnected, allowing the individual to maintain a high degree of autonomy. Using an exoskeleton to shape the robots' morphology, the swarm can promote cohesion-through-collision at the phototactic destination. We find that the steady state of a successful strategy can give an insight to the ability of the learning process to converge. |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700