A Workshop on
Fully sponsored by IEEE RAS TC on Biorobotics
 

Invited Papers




CV

Hilary BART-SMITH
Bart-Smith Labs, VA, USA
“Title: Towards a Mission Configurable Stealth Underwater Batoid”
Time: 15:00-15:15
Abstract: For millions of years, aquatic species have utilized the principles of unsteady hydrodynamics for propulsion and maneuvering. They have evolved high endurance swimming capabilities that can outperform current underwater vehicle technology in the areas of stealth, maneuverability and control authority. As such, they provide a rich source for inspiration in the development of the next generation of unmanned underwater vehicles (UUVs). Batoid fishes, including the manta ray, Manta birostris, the cownose ray, Rhinoptera bonasus, and the Atlantic stingray, Dasyatis sabina, have been identified as a high performing species due to their ability to migrate long distances, maneuver in spaces the size of their tip-to-tip wing span, produce enough thrust to leap out of the water, populate many underwater regions, and attain sustained swimming speeds of 2.8 m/s with low flapping/undulating frequencies. These characteristics make them an ideal platform to emulate in the design of a bio-inspired UUV. The current study is focused on trying to understand the biological principles for locomotion. Specifically, how the coupled spanwise curvature and chordwise traveling wave of enlarged pectoral fins of the ray enable thrust and maneuvering. Direct comparisons between observed swimming motions and the underlying cartilage structure of the pectoral fin have been made. A simple yet powerful analytical model to describe the swimming motions of batoid fishes has been developed and is being used, in conjunction with in-house computational fluid dynamics, to quantify their hydrodynamic performance. This model is also being used as the design target for artificial pectoral fin design. Experimentally, active tensegrity structures are being used to replicate the observed motion of the biological pectoral fin. These are being used to understand the relationship between form and swimming function of batoid fishes. Using the knowledge gained through the combined experimental/theoretical studies, a robotic manta ray, MantaBot, has been built and tested.



CV

Kyujin CHO
Seoul National University, Korea
“Title: Increasing the thrust of a robotic dolphin using a variable stiffness flapping mechanism”
Time: 14:45-15:00
Abstract: Compliant fins have been used by many underwater robots with flapping mechanisms since they are known to produce larger thrust more effectively. However, the question of “what is the optimal stiffness of a compliant fin”, which is important for designing a compliant fin based flapping mechanism for underwater robots is yet to be answered. In this talk, I will first present a guideline that could be used to optimize the stiffness of a compliant fin, termed “half-pi phase delay condition”. This guideline can be used to experimentally find the optimal stiffness of a flapping fin for generating maximum thrust under certain flapping frequency. As the frequency increases, larger forces will be applied to the fins and the optimal stiffness will increase. Therefore, a mechanism that could actively vary the stiffness of a flapping mechanism to maximize the thrust under various flapping frequencies is necessary. I will present a novel Variable Stiffness Flapping mechanism (VaSF) inspired by the anatomy of the fluke, which consists of rigid and compliant discs alternatively connected in series. Two tendons driven by servo motors control the stiffness of the mechanism. Experimental results show that the thrust generated by the flapping mechanism can be maximized by varying its stiffness under various conditions. Using the VaSF mechanism, the stiffness of the peduncle of a robotic dolphin is actively controlled and the thrust of the robotic dolphin is shown to increase by simply changing the stiffness of its flapping mechanism.



CV

Jonathan CLARK
FAMU/FSU College of Engineering, Florida, USA
“Title: Dynamic Templates and Rapid, Multi-Modal Legged Locomotion”
Time: 09:50-10:05
Abstract: Finely tuned legged systems that explicitly exploit their body’s natural dynamics have begun to rival specific performance criteria, such as speed over smooth terrain, of the most accomplished biological systems. The success of these robots is, in part, due to the insights into the underlying dynamics of motion lent by the study of reduced-order models or “templates”. Relevant models exist for a number of modalities of motion including: sagittal plane walking and running (SLIP), lateral plane running (LLS) and climbing (FG). These models have been shown to match biological data, and have guided the mechanical and control design for rapidly running and climbing robots. We hypothesize that for robots to move fast and efficiently in multiple domains or via modes of operation they should be able to selectively anchor multiple templates. This talk will address some challenges associated with instantiating more than one model or mode of locomotion including altering the geometry, actuation and passive compliance of the limbs. Insights from our robots that can run and climb (SCARAB) and climb and glide (ICAROS) will be used to inform this discussion.



CV

Dario FLOREANO
EPFL, Switzerland
“Title: Multi-modal Flying Robots”
Time: 16:15-16:30
Abstract: I will present an overview of my lab's efforts to develop autonomous robots capable of flying in cluttered environments and in safe interaction with humans. I will start by presenting miniature and small-size robots capable of performing collision-free flight and altitude control indoor and outdoor by means of insect-inspired vision and control. I will then describe transition from flying to walking locomotion, and I will conclude with a description of current efforts to develop flying robots capable of surviving and exploiting collisions, just like insects do, in order to explore semi-collapsed buildings or extremely cluttered environments.



CV

Fumiya IIDA
ETH, Zurich, Switzerland
“Title: Unconventional Actuation Principles for Multi-Gaited Legged Robot Locomotion”
Time: 11:20-11:35
Abstract: Despite the intensive studies on animals' locomotion, it has not been fully clarified the underlying mechanisms and principles of gait patterns observed in different species in various environments. Because of the lack of our biological knowledge, most of the legged robots in the past could only mimic the gait appearance of animals such as kinematics of leg joints or duty ratios of stance phases, but the other seemingly more fundamental aspects (such as energy efficiency and constraints in musculoskeletal dynamics) have not been studied systematically. In order to gain additional insights into the nature of multi-gait legged locomotion from the bio-inspired robotics standpoint, we would like to discuss in our presentation three distinctive approaches in which energetically efficient legged robots can be designed and analyzed. First, we explain the influence of body morphology (e.g. leg length and body length in a quadruped locomotion) in the multi-gait locomotion, where we developed a series of robots with different morphologies to analyze the relationship between morphologies, resonance frequencies and emerging gait patterns. Second, we introduce a set of gait patterns generated from unconventional actuators called Linear Multi-Modal Actuators, which elaborates the potential roles of muscles for generating different gait patterns. And third, we also argue the importance of sensory-motor learning in the context of emergence of gait patterns through our recent studies on computational neuroscience on adaptive legged locomotion. These case studies will then be summarized into a few challenges and perspectives for the future studies on multi-gait legged locomotion.



CV

Sangbae KIM
MIT, USA
“Title: Galloping Control for Quadruped Robots: Application to the MIT Cheetah Robot”
Time: 10:20-10:35
Abstract: In this research, we seek a controller design scheme that allows robust and variable speed galloping gaits of quadrupedal robots. Update on the MIT Cheetah project will be also presented.
For more information, please see the attached document: Abstract



CV

Kyoungchul KONG
Sogang University, Seoul, Korea
“Title: Different Approaches in the Development of Bio-inspired Locomotive Robots”
Time: 10:05-10:20
Abstract: Many approaches have been introduced to improve the gait stability and the locomotion speed of legged robots. A simple, yet the most effective, way is to learn from the nature, i.e., the stable and dexterous locomotion of animals. In this talk, the locomotion of fast running animals (e.g., American Akita dogs) is analyzed, and its mathematical model is introduced for the design and control of legged robots. In particular, we introduce three different approaches to effectively realize the locomotion of animals: 1) mechanical design approach, 2) control approach, and 3) animal-robotization approach. From these aspects, we have developed three different bio-inspired robots, called Cheetaroid series. The Cheetaroid-I is a fully actuated high-speed running robot with muscle-like linear actuators. The bio-inspired gait-pattern generation method and the associated feedback control method are introduced in this talk. The Cheetaroid-II, the gait motions of which are generated by mechanical link systems, is also introduced in this presentation. This robot has four legs that propel the robot body by mimicking the leg motions of an animal running fast. For the easiness in control and the minimal weight of the overall robot system, all the four legs are connected by mechanical links and actuated by only one actuator. In addition to the actuated leg joints, the proposed robot also has compliant feet for enhancing gait efficiency. The parameters of the mechanical link system are selected by an optimization process such that the simulated link motion is close to the leg motions of the animal. The Cheetaroid-III is actually a dog itself, the behavior of which is controlled (that is, “robotized”) by distracting cognition of the animal by virtual reality.



CV

Mirko KOVAC
Imperial College London, UK
“Title: Bio-inspired design: A perspective for multi-modal flying robots”
Time: 16:00-16:10
Abstract: Biologically inspired design is an engineering synthesis approach for artificial systems that are conceptually inspired from biology. Its premise is the creation of unconventional mechanical solutions that have the potential to outperform classical engineering designs. In the field of flying micro robots, bioinspired design principles such as flapping wing flight or optical flow based navigation are often employed but it is not always clear where exactly and how the biological systems act as a source of inspiration. In this talk, I will formalize successful bioinspired design strategies in the perspective of flying robot development. For illustration, I will present a number of bio-inspired robotic case studies that focus on multi-modal locomotion that can greatly improve the mobile capabilities of micro robots in rough terrain.



CV

Kin Huat LOW
Nanyang Technological University, Singapore
“Title: A Bio-inspired Perching with Unmanned Aerial Vehicles”
Time: 16:10-16:15
Abstract: Unmanned Aerial Vehicles (UAVs) have been applied more and more in the field of surveillance, such as traffic monitoring, search and rescue, and military reconnaissance. For such UAVs, the most critical problem is the endurance. Current UAVs have to keep flying during the surveillance missions, decreasing the efficiency of energy consumption dramatically. This is where the concept of perching applies. The concept of perching is apparently derived from the master of flight in nature–birds. Unlike current UAVs, birds perch now and then in some strategic locations where they can continue the surveillance on their domain for predator and prey while conserving energy for more critical needs. It is hoped that the bio-inspired perching provides a promising solution to the endurance issue of UAVs. The present research work aims at developing a systematic methodology for intelligent perching with UAV platforms based on inspirations from birds. Such a bio-inspired methodology can readily provide an effective solution for perching a UAV autonomously and reliably to some targets. By intelligent perching it means the approaching of the target can be sensed and corresponding maneuvers can be performed to secure the UAV platform to the perch. What’s more, the orientation of the UAV can be adjusted if misalignment or slippage occurs. Moreover, the applicable UAV platforms may include copters with multi-rotors, fixed wing UAVs and even robotic birds with flapping wings, and the targets of perching taken into account comprehend tree brunches, walls, ground, and roofs which birds usually perch to. Such a broad range of perching circumstances is in order to generalize the methodology of intelligent perching to more potential applications. Other topics in this talk include a comprehensive methodology of bio-inspired perching for UAVs ranging from bio-inspirations from birds to perching mechanism development, and the control scheme for autonomous perching as well.



CV

Matt SPENKO
Illinois institute of technology, USA
“Title: Designing multimodal mobile robots: translating complex biology into simple mechanical structures”
Time: 11:50-12:05
Abstract: Many researchers use bio-inspired design principles to create mobile robots with the ability to locomote in multiple domains (i.e. terrestrial, aerial, scansorial, or aquatic locomotion). Despite these efforts, few, if any, robots can move with the grace of animals in even one environmental domain, much less two or more. We argue that this results from an inherent correlation between versatility and mechanical complexity in mobile robot design, a correlation that does not necessarily exist in animals. In other words, in order to create a more versatile mobile robot, the most common approach is to add actuators and sensors. However, this increases system mass and control complexity, which typically degrades performance. This paper examines the aforementioned correlation and identifies several designs that are the exceptions to the rule in an effort to meld bio-inspired design principles with current technological limitations.



CV

Hiroto TANAKA
Chiba University, Japan
“Title: Passive Dynamic Flying Based on Flapping Wings in Nature”
Time: 16:30-16:45
Abstract: Flapping flight of natural flyers such as insects or small birds is promising source of inspiration for small flying robots. In small-scale aircrafts with low Reynolds number of less than 104, flapping wings could generate greater aerodynamic forces or achieve higher efficiency than conventional fixed wings by utilizing unsteady vortices. The wing kinematics of the flapping wings of insects and animals, however, is complicated: The wing motion is composed of flapping, feathering and elevation (or lead-lag motion), and the flapping frequency ranges 100 to 102 Hz. Those are difficult to directly mimic in robotic flapper with the severe limitation of the gross weight. Here, I propose a strategy to reduce the complexity of the active wing kinematics by exploiting passive motions with flexible structures. In this approach, the flapping motion is actively produced with an actuator, and the feathering motion is passively achieved with the flexible wingbase or the flexible wing membranes. Since the angle of attack of the wing, the angle between the wing chord and incoming flow direction, directly changes with the feathering angle, it is crucial to create appropriate feathering motion for high aerodynamic performance. On the other hand, elevation or lead-lag motion can be neglected. In this talk, three examples of the simplified biologically-inspired flapping mechanism with flexible structures based on butterflies, hoverflies and hummingbirds will be introduced. In addition, micro fabrication methods used for those insect-scale flexible structures will be explained. In forward flapping flight of large lepidopteran like swallowtail butterflies, active feathering is relatively small. Based on this observation, a tailless ornithopter having the same morphology of the swallowtail butterfly was created [1]. The ornithopter demonstrated that stable forward flight is possible only with the simple flapping wings without active feathering. The MEMS-veined wings also revealed that change of the wing stiffness caused by variation of the venation pattern lead to different flight trajectory and flight speed. In contrast to the lepidopteran, dipteran wings show remarkable feathering motion at the wingbase. A Dipteran-based tethered flapping mechanism was used to investigate the effect of the passive feathering at the wingbase on lift generation in hovering flight [2]. It was found that the flexible hinge at the wingbase with an appropriate stiffness produces proper passive feathering similar to the model hoverfly, generating almost the same lift as the weight of the hoverfly even with the rigid wing surface. As for hummingbirds’ hovering flight, we focused on the deformation of the wing surface. We measured the wing deformation of a hovering hummingbird with high-speed video cameras, then tested our hypothesis that similar wing deformation can be realized with a flexible film membrane supported with stiff leading-edge and body-side frames using tethered mechanical flapper. As a result, the film membrane twisted like the hummingbird’s wing in upstroke and generated almost the same vertical force as the measured hummingbird. Those result suggested that main aerodynamic features of the complex wing kinematics in nature could be reconstructed with the simplified mechanical flapping with passive feathering. In turn, it implies that natural flyers could utilize passive wing motion or deformation to some extent.



CV

James TANGORRA
Drexel University, USA
“Title: Multi-modal swimming: multiple fins for many gaits”
Time: 14:30-14:45
Abstract: Bony fish coordinate the motions of the body and of multiple fins and to produce the forces used for swimming and maneuvering. Low speed propulsion and maneuvers are often driven by the pectoral fins, and as speeds increase the pectoral fins’ motions are supplemented by motions of the peduncle, caudal, dorsal, and anal fins. High speed swimming and high accelerations are often driven by the body and the caudal fin alone. Backward swimming and braking, which occur when swimming through occluded environments, will frequently involve all seven of the fish’s flexible fins. In this work we combine studies of the Bluegill sunfish (Lepomis macrochirus) and of biorobotic models of the fish’s peduncle, and dorsal, caudal, and anal fins to investigate forward and backward swimming. Kinematic patterns used by the biological fish were implemented in the biorobotic models, and the impact of individual fins and the effect of phase relations among fins on the magnitude and direction of propulsive force were evaluated. Results indicate that the dorsal and anal fins, which are often omitted in human engineered, bio-inspired vehicles, significantly influence the magnitude and direction of forces. Also, the correct orchestration of the dorsal, anal, and caudal fin is critical to backward swimming which employs very different kinematics than used in forward swimming.



CV

Florentin WORGOTTER
Bernstein Center for Computational Neuroscience, Germany
“Title: Interaction learning for dynamic movement primitives used in cooperative robotic tasks”
Time: 11:35-11:50
Abstract: Since several years dynamic movement primitives (DMPs) are more and more getting into the center of interest for flexible movement control in robotics. In this study we introduce sensory feedback together with a predictive neuronal learning mechanism which allows tightly coupled dual-agent systems to learn adaptive, sensor-driven interaction based on DMPs. The coupled conventional (no-sensors, no learning) DMP-system automatically equilibrates and can still be solved analytically allowing to derive conditions for stability. When adding adaptive neuronal sensor control we can show that both agents learn to cooperate. Interestingly, all these mechanisms are entirely based on low level interactions without any planning or cognitive component.




NANYANG UPEC-LISSI