Tuesday, June 23
09:00 - 10:00 | Keynote:
George Lauder, Organismic and Evolutionary Biology, Harvard University
Robotic models of aquatic locomotion have many advantages over studying live animals, including the ability to manipulate and control individual morphological or kinematic factors that affect performance, substantially easier measurement of locomotor forces and torques, and the ability to abstract complex organismal designs into simpler components. Such simplifications, while not without their drawbacks, facilitate interpretation of how individual traits alter swimming performance and the discovery of underlying physical principles. In this presentation I will discuss the use of a variety of robotic models for fish locomotion, ranging from simple flexing panels to complex biomimetic designs incorporating flexible, actively moved, fin rays on multiple fins. Mechanical devices have provided great insight into the dynamics of aquatic propulsion and, integrated with studies of locomotion in freely-swimming fishes, promise new insights into how fishes move through the water.
10:00 - 10:20 | Break / Refeshments
10:20 - 12:00 | Session 2A
Malcolm MacIver, Mechanical Engineering, Northwestern University
"Electric Fish Robotics"
Weakly electric fish are a popular model system within sensory neurobiology and, more recently, biomechanics. Over the past 60 years since their discovery, approximately three thousand studies have been published, encompassing ecology and evolution, behavior, neurophysiology, and anatomy. Historically the field has attracted quantitatively oriented scientists with training in physics or engineering, which makes it an ideal system for the building of bio-inspired machines, with that endeavor’s high demands on quantitative detail. This talk will survey what building machines based on electric fish has contributed in terms of science and technology, and take stock of what this suggests about where building and discovery can most usefully intersect.
Kiisa Nishikawa, Center for Bioengineering Innovation, Northern Arizona University
"A neuromuscular algorithm for a powered foot-ankle prosthesis shows robust control of level walking and stair ascent"
The iWalk BiOM is a powered, foot-ankle prosthesis for persons with trans-tibial amputation. Provision of motor power permits faster walking but raises the issue of control. Although the BiOM performs well across a range of level and ramp walking speeds, more robust control algorithms could improve all-terrain walking. We developed a “winding filament” hypothesis (WFH) for muscle contraction that incorporates a role for the giant titin protein. Our goal was to develop a WFH-based control algorithm for the BiOM prosthesis. The algorithm incorporates a pair of virtual muscles: a dorsiflexor and a plantarflexor. The algorithm estimates the torque produced by each muscle based on its length and level of activation. The dorsiflexor is activated at 50% of maximum force during swing, and the plantarflexor is activated at 50% of maximum force during stance. We tested subjects during level walking, stair ascent, descent, and backwards walking. During level walking, the WFH-based algorithm produces torques similar to the BiOM stock controller and human ankle. The algorithm also produces normal ankle torques during stair ascent, with minimal sensing (i.e., ankle angle) and no change in activation or model parameters.
Emily Standen, Comparative and Evolutionary Biomechanics, University of Ottawa
"What makes a fin useful for walking?"
Polypterus is a basal ray-finned fish that can locomote effectively in water and on land. How these fish use their bodies and fins for propulsion differs between environments. Of particular interest is the ability of fish to use both the medial and lateral fin surface for support throughout stance during terrestrial locomotion. While it is not clear if the variation in fin motion represents intentional behaviour or if it is an ‘accident’ of body weight shifting, it is without question that the complexity of the fin motion and the forces associated with the extreme fin bending pose interesting control, sensory and functional challenges. Our overarching goal is to understand the control and functional flexibility of Polypterus fins and body when used in novel environments. More specifically we want to address how bones at the base of fin rays tolerate extreme ranges of motion and also how to optimize body wave form and ground contact to maximize walking performance. Answering these questions will tell us if fish are maximizing their locomotory performance in terrestrial environments and may provide mechanical and control modulation insights for unique biomimetic systems.
Auke Ijspeert, Biorobotics Laboratory, École Polytechnique Fédérale de Lausanne EPFL
"Neuromechanical models of locomotion: from biology to robotics"
The ability to efficiently move in complex environments is a fundamental property both for animals and for robots, and the problem of locomotion and movement control is an area in which neuroscience and robotics can fruitfully interact. Animal locomotion control is in a large part based on spinal cord circuits that combine reflex loops and central pattern generators (CPGs), i.e. neural networks capable of producing complex rhythmic or discrete patterns while being activated and modulated by relatively simple control signals. These networks are located in the spinal cord for vertebrate animals and interact with the musculoskeletal system to provide "motor primitives" for higher parts of the brain, i.e. building blocks of motor control that can be activated and combined to generate rich movements. In this talk, I will present how we model the spinal cord circuits of lower vertebrates (lamprey and salamander) using systems of coupled oscillators, and how we test these models on board of amphibious robots. The models and robots were instrumental in testing some novel hypotheses concerning the mechanisms of gait transition, sensory feedback integration, and generation of rich motor skills in vertebrate animals. I will also discuss how the models can be extended to control biped locomotion, and how they can help deciphering the respective roles of pattern generation, reflex loops, and descending modulation in human locomotion.
Madhusudhan Venkadesan, Mechanical Engineering and Materials Science, Yale University
" Stiffness of the human foot and evolution of the transverse arch"
Sustained walking and running in humans is made possible by our stiff propulsive feet, unlike the flexible feet of other primates. Morphologically, human feet are unique among primates in having two distinct arches, the longitudinal and the transversal arches. Human ancestors had feet with a longitudinal arch well over 4 million years ago, but the transversal arch emerges in the fossil record only 3.2 million years ago, coincident with regular bipedal walking and running. We wondered whether and how the uniquely arched structure of the human foot affects our locomotive abilities. Through experiments and mathematical modeling, we show that the transversal arch is almost entirely responsible for maintaining a stiff foot during propulsion, i.e. stiffness against bending in the sagittal plane. Contrary to existing theories, we show that the longitudinal arch and the elastic plantar fascia have a negligible effect on the bending stiffness. The source of this additional stiffness is because the transversal arch couples longitudinal bending in the sagittal plane to stretching of internal connective tissue in the transversal direction. Using the length-wise torsion of the metatarsal bones, we estimate the curvature of the transversal arch in humans, chimpanzees, the feet of the 3.2 million year old Australopithecus afarensis, and the 1.8 million year old OH-8 fossil. Unlike the chimpanzee, the transversal arches of humans, A.afarensis and the OH-8 feet are all better suited for maintaining stiff feet. The transverse arch may therefore be central to the evolution of bipedality. We also suggest that impairment and recovery of foot function in humans may be better predicted by the transversal arch curvature rather than the longitudinal arch that is currently used as a surrogate of healthy feet. Transversal arches would also benefit the design of robotic and prosthetic feet by allowing them to remain light yet sufficiently stiff.
12:00 - 13:15 | Lunch
13:20 - 14:50 | Session 2B
Russ Tedrake, Electrical Engineering and Computer Science, MIT
"High-speed flight through forests using onboard stereo vision"
Over the last 5 years we have been collaborating with Andy Biewener at Harvard and David Lentink at Stanford to learn what we can about high-speed maneuvering flight in birds. Their experiments with instrumented birds have led us to try to match their performance using a custom maneuverable fixed-wing aircraft with onboard stereo cameras flying through obstacles at 7-10 mps. Relative to the birds, our planes take a very model-based approach with onboard stereo vision computing realtime geometric estimates of the obstacles and with online feedback motion planning and robustness verification. I will describe our results to date along with the power (and limitations) of this approach.
Koushil Sreenath, Mechanical Engineering, Carnegie Mellon University
"Dynamic Aerial Manipulation in Birds-of-Prey and Aerial Robots"
In this talk, I will present the design of planning and control policies for achieving dynamic aerial manipulation. I will discuss a method that first draws inspiration from aerial hunting by birds of prey that enables a micro aerial vehicle with an actuated appendage to perform grasping and object retrieval at high speeds. I will also present an initial comparison of the dynamic grasping maneuver between the micro aerial vehicle and the bird of prey. I will next show how a coordinate-free, geometric formulation of the dynamics of a quadrotor carrying a suspended payload allows us to synthesize nonlinear geometric controllers with with the ability to recover from almost any global state. Finally, I will present the problem of cooperative transportation of a cable-suspended payload from multiple aerial vehicles, and show how we can design dynamically feasible trajectories that can handle hybrid dynamics resulting from the cable tension going to zero.
Ioannis Poulakakis, Mechanical Engineering, University of Delaware
"Canonical Models for Legged Locomotion Across Scales"
Macroscopically, legged locomotion can be understood through reductive canonical models -- often termed templates -- the purpose of which is to capture the dominant features of a locomotion behavior without delving into the fine details of a robot’s (or animal’s) structure and morphology. Such models offer unifying, platform-independent, descriptions of task-level behaviors, and inform control design for legged robots. This talk will discuss reductive locomotion models for diverse legged systems, ranging from slow-moving, palm-size, eight-legged crawlers to larger quadrupeds, and will focus on the role of such models in capturing gait energetics in quadrupeds as well as in integrating locomotion control and motion planning for miniature robots.
Huai-Ti Lin, HHMI Janealia Research Campus, USA
"Dragonfly behavioral strategies for prey interception"
The dragonfly is a superb aerial predator that intercepts flying insects in a fraction of a second guided by vision alone. Only recently, we were able to accurately track the head orientation in 3D and reconstruct the visual experience of a foraging dragonfly. One challenge for dragonfly vision during this behavior is to quickly identify a “suitable prey” within catchable range and speed. Intuitively, this involves target distance estimation. However, our data support a simpler solution, where a heuristic template of “suitable prey” is necessary and sufficient. We have further found a neural substrate that could account for such template.
Christian Hubicki, Dynamic Robotics Lab, Oregon State University
"Toward Controlling a Bipedal Robot to Run like a Bird: Combining Robot Mechanics and Animal Objectives"
Ground-running birds are adept bipedal runners. From quail to guinea fowl to ostrich, they are capable of negotiating obstacles of varying height and ground impedance. We seek to imbue our bipedal robots with similar performance. We hypothesize that a robot can be made to run like a bird if given sufficiently analogous passive dynamics and controlled to respect a practical set of control objectives. We present experiments probing the control objectives of cursorial birds, numerical studies of spring-mass robot models running with these objectives, and ongoing robot experiments working toward achieving avian obstacle negotiation on the bipedal robot, ATRIAS.
14:50 - 16:30 | Refeshments/ Poster Session 2 (Even Numbered Submissions)
16:30 - 17:50 | Session 2C
Katie Byl, Mechanical Engineering, Univerisity of California, Santa Barbara
"Limbed Mobility with RoboSimian"
RoboSimian is a quadruped designed by JPL to complete in the DARPA Robotics Challenge (DRC). Given an emphasis within the DRC on tasks requiring manipulation of devices (e.g., tools) designed for human operation, a humanoid form seems a more natural winning solution. On the other hand, a four-legged design arguably provides greater robustness for mobility in situations with variable environments and noisy perception. Each of RoboSimian's limbs is designed with seven identical actuators, arranged in a kinematic chain, toward increased dexterity and strength. By contrast, a quadruped designed for speed and efficiency of locomotion would typically use fewer actuators, with care taken to reduce the mass and inertia that must be swung when each leg moves in free space. This talk will discuss the benefits, challenges, and lesson learned during development and testing by JPL and UCSB of control and planning algorithms specifically for mobility within the DRC.
Andrew Spence, Department of Bioengineering, Temple University
"Neuromechanics and optogenetics: have the tools to precisely dissect the neural and mechanical contributions to locomotion in intact, freely behaving animals arrived?"
Animal locomotion is an integrative phenomenon. Neural, musculoskeletal, body and environmental dynamical systems interact to produce it. Whilst our holistic descriptions of intact moving animals are providing increasing levels of insight (e.g. dynamical systems models of movement kinematics), our ability to perturb and manipulate the “subsystems” (neural, musculoskeletal, etc.) involved in locomotion is improving in dramatic ways. Optogenetics makes possible specific, causal, reversible, fast manipulation of the nervous system; closed loop experimental techniques are making possible precise, repeatable mechanical perturbations. This talk will present the state of the art of optogenetics, elicit a discussion of how it may be best applied to movement science, and describe the potential for it to answer long-standing questions in locomotion. Recent work that targets two such questions and is designed to employ these new technologies will be presented, namely: 1) how have ultimate constraints such as energetic cost and stability shaped the control of gait? and 2) how is proprioceptive sensory feedback used during fast locomotion.
Koh Hosoda, Adaptive Machine Systems, Osaka University
"Function of the Foot Complex for Walking"
Our foot is composed from 26 bones, many ligaments, and tendons. Its structure is compliant and very complicated. During walking, the foot touches the ground, supports the body weight, negotiates with ground, and supplies propulsion force. During this process, the bones move quite amount, and the relative motion will provide function of foot. In this presentation, I will talk about our approach to observe and understand the process, by utilizing a gait simulator and cadaver foot, and a walking robot.
Steve Collins, Mechanical engineering, Carnegie Mellon University
"Optimizing (artificial) ankle function during walking"
Animal locomotion is complex, and the space of possible motor patterns is vast. Subtle changes in activity have cascading effects, often strongly affecting overall performance. Performance itself can be difficult to observe due to variability and long mechanical, biochemical and neural delays.How do animals discover effective coordination strategies given these challenges? Optimization of robotic devices that interact with humans during gait may provide insights. We will discuss strategies for optimizing wearable robot function, both online and offline, using a variety of algorithms and objective functions. Initial results with ankle prostheses and exoskeletons suggest that performance for individual humans can be improved, given a well-posed problem and sufficient time, and suggest strengths and weaknesses of various candidate approaches to optimizing animal locomotion.
18:30 - 19:30 | Planery talk : Stata Center 32-123
Robert J. Full, Integrative Biology, University of California, Berkeley
"Motion Science of Animals and Machines – An Exemplar of Convergence"
The motion science of animals and machines is an exemplar of convergence integrating knowledge, tools, and ways of thinking and interacting from biology, physics, applied mathematics, and engineering to form a comprehensive synthetic framework for addressing scientific and societal challenges that exist at the interfaces of many traditional disciplines. Advancements embrace the challenges inherent in the approaches to biological inspiration, physical modeling, optimality versus satisficing assumptions, theory versus empirical measurements, spatial and temporal variation, simple versus representative models, selection of model animals or extreme performers, use of enabling technologies, and effective infrastructures for collaboration among disciplines, universities and businesses. In part, our admiration of animals relates to their robustness - the ability to withstand perturbations in structure without change in function - and includes concepts such as modularity, redundancy, hierarchical and heterarchical organizational structures, damage resistant, fail-safe and fault tolerant designs, self-repair, learning, adaptation, anticipation and creativity. Although biology offers enormous potential for novel designs, we must realize that natural selection is not engineering and nature is severely constrained by evolution, development, multi-functionality and sexual selection. Increasingly, key enablers, novel organizational structures, and synergies with businesses will determine the rate and direction of motion science in the future. The novel science, mathematics, and robots emerging from motion science will continue to lead to new relationships of natural and human technologies.