Videos: Autonomous Car Drifting, Aquatic Drone, and Jet-Powered Robot

Video Friday is your weekly collection of superior robotics movies, collected by your folks at IEEE Spectrum robotics. This week, we’re that includes a particular collection of movies from ICRA 2023! We additionally put up a weekly calendar of upcoming robotics occasions for the subsequent few months. Please ship us your occasions for inclusion.

Energy Drone & Robotics Summit: 10–12 June 2023, HOUSTON, TEXAS, USA

RoboCup 2023: 4–10 July 2023, BORDEAUX, FRANCE

RSS 2023: 10–14 July 2023, DAEGU, SOUTH KOREA

IEEE RO-MAN 2023: 28–31 August 2023, BUSAN, SOUTH KOREA

IROS 2023: 1–5 October 2023, DETROIT, MICHIGAN, USA

CLAWAR 2023: 2–4 October 2023, FLORIANOPOLIS, BRAZIL

Humanoids 2023: 12–14 December 2023, AUSTIN, TEXAS, USA

Enjoy in the present day’s movies!

“Autonomous Drifting With 3 Minutes of Data Via Learned Tire Models,” by Franck Djeumou, Jonathan Y.M. Goh, Ufuk Topcu, and Avinash Balachandran from University of Texas at Austin, and Toyota Research Institute, in Los Altos, Calif.

Abstract: Near the boundaries of adhesion, the forces generated by a tire are nonlinear and intricately coupled. Efficient and correct modeling on this area might enhance security, particularly in emergency conditions the place excessive forces are required. To this finish, we suggest a novel household of tire power fashions based mostly on neural atypical differential equations and a neural-ExpTanh parameterization. These fashions are designed to fulfill bodily insightful assumptions whereas additionally having enough constancy to seize higher-order results straight from car state measurements. They are used as drop-in replacements for an analytical brush tire mannequin in an current nonlinear mannequin predictive management framework. Experiments with a custom-made Toyota Supra present that scarce quantities of driving information—lower than 3 minutes—is enough to realize high-performance autonomous drifting on varied trajectories with speeds as much as 45 miles per hour. Comparisons with the benchmark mannequin present a 4x enchancment in monitoring efficiency, smoother management inputs, and quicker and extra constant computation time.

“TJ-FlyingFish: Design and Implementation of an Aerial-Aquatic Quadrotor With Tiltable Propulsion Units,” by Xuchen Liu, Minghao Dou, Dongyue Huang, Songqun Gao, Ruixin Yan, Biao Wang, Jinqiang Cui, Qinyuan Ren, Lihua Dou, Zhi Gao, Jie Chen, and Ben M. Chen from Shanghai Research Institute for Intelligent Autonomous Systems, Tongji University, Shanghai, China; Chinese University of Hong Kong; Nanjing University of Aeronautics and Astronautics, Nanjing, Jiangsu, China; Peng Cheng Laboratory, Shenzhen, Guangdong, China; Zhejiang University, Hangzhou, Zhejiang, China; Beijing Institute of Technology, in China; and Wuhan University, Wuhan, Hubei, China.

Abstract: Aerial-aquatic automobiles are able to transferring within the two most dominant fluids, making them extra promising for a variety of functions. We suggest a prototype with particular designs for propulsion and thruster configuration to deal with the huge variations within the fluid properties of water and air. For propulsion, the working vary is switched for the completely different mediums by the dual-speed propulsion unit, offering enough thrust and likewise making certain output effectivity. For thruster configuration, thrust vectoring is realized by the rotation of the propulsion unit across the mount arm, thus enhancing the underwater maneuverability. This paper presents a quadrotor prototype of this idea and the design particulars and realization in apply.

“Towards Safe Landing of Falling Quadruped Robots Using a 3-DoF Morphable Inertial Tail,” by Yunxi Tang, Jiajun An, Xiangyu Chu, Shengzhi Wang, Ching Yan Wong, and Ok. W. Samuel Au from the Chinese University of Hong Kong and the Multiscale Medical Robotics Centre, in Hong Kong.

Abstract: Falling cat drawback is well-known the place cats present their tremendous aerial reorientation functionality and may land safely. For their robotic counterparts, the same falling quadruped robotic drawback has not been totally addressed, though reaching as protected a touchdown because the cats has been more and more investigated. Unlike imposing the burden on touchdown management, we strategy to protected touchdown of falling quadruped robots by efficient flight section management. Different from current work like swinging legs and attaching response wheels or easy tails, we suggest to deploy a 3-DoF morphable inertial tail on a medium-size quadruped robotic. In the flight section, the tail with its most size can self-right the physique orientation in 3D successfully; earlier than landing, the tail size might be retracted to about 1/4 of its most for impressing the tail’s facet impact on touchdown. To allow aerial reorientation for protected touchdown within the quadruped robots, we design a management structure that’s verified in a high-fidelity physics simulation surroundings with completely different preliminary situations. Experimental outcomes on a custom-made flight-phase check platform with comparable inertial properties are offered and present the tail’s effectiveness on 3D physique reorientation and its quick retractability earlier than landing. An preliminary falling quadruped robotic experiment is proven, the place the robotic Unitree A1 with the 3-DoF tail can land safely topic to non-negligible preliminary physique angles.

“Nonlinear Model Predictive Control of a 3D Hopping Robot: Leveraging Lie Group Integrators for Dynamically Stable Behaviors,” by Noel Csomay-Shanklin, Victor D. Dorobantu, and Aaron D. Ames from Caltech, in Pasadena, Calif.

Abstract: Achieving steady hopping has been a trademark problem within the area of dynamic legged locomotion. Controlled hopping is notably tough as a consequence of prolonged durations of underactuation mixed with very quick floor phases whereby floor interactions should be modulated to control a world state. In this work, we discover the usage of hybrid nonlinear mannequin predictive management paired with a low-level suggestions controller in a multirate hierarchy to realize dynamically steady motions on a novel 3D hopping robotic. In order to display richer behaviors on the manifold of rotations, each the planning and suggestions layers should be designed in a geometrically constant vogue; subsequently, we develop the mandatory instruments to make use of Lie group integrators and acceptable suggestions controllers. We experimentally display steady 3D hopping on a novel robotic, in addition to trajectory monitoring and flipping in simulation.

“Fast Untethered Soft Robotic Crawler with Elastic Instability,” by Zechen Xiong, Yufeng Su, and Hod Lipson from Columbia University, New York, N.Y.

Abstract: Enlightened by the fast-running gait of mammals like cheetahs and wolves, we design and fabricate a single-actuated untethered compliant robotic that’s able to galloping at a velocity of 313 millimeters per second or 1.56 physique lengths per second (BL/s), quicker than most reported gentle crawlers in mm/s and BL/s. An in-plane prestressed hair clip mechanism (HCM) made up of semirigid supplies, i.e., plastics are used because the supporting chassis, the compliant backbone, and the power amplifier of the robotic on the identical time, enabling the robotic to be easy, speedy, and powerful. With experiments, we discover that the HCM robotic locomotion velocity is linearly associated to actuation frequencies and substrate friction variations aside from concrete surfaces, that tethering slows down the crawler, and that uneven actuation creates a brand new galloping gait. This paper demonstrates the potential of HCM-based gentle robots.

“Nature Inspired Machine Intelligence from Animals to Robots,” by Thirawat Chuthong, Wasuthorn Ausrivong, Binggwong Leung, Jettanan Homchanthanakul, Nopparada Mingchinda, and Poramate Manoonpong from Vidyasirimedhi Institute of Science and Technology (VISTEC), Thailand, and the Maersk Mc-Kinney Moller Institute, University of Southern Denmark.

Abstract: In nature, residing creatures present versatile behaviors. They can transfer on varied terrains and carry out spectacular object manipulation/transportation utilizing their legs. Inspired by their morphologies and management methods, we now have developed bioinspired robots and adaptive modular neural management. In this video, we display our 5 bioinspired robots in our robotic zoo setup. Inchworm-inspired robots with two electromagnetic ft (Freelander-02 and AVIS) can adaptively crawl and stability on horizontal and vertical steel pipes. With particular design, the Freelander-02 robotic can adapt its posture to crawl beneath an impediment, whereas the AVIS robotic can step over a flange. A millipede-inspired robotic with a number of physique segments (Freelander-08) can proactively adapt its physique joints to effectively navigate on bump terrain. A dung beetle–impressed robotic (ALPHA) can transport an object by greedy it with its hind legs and on the identical time stroll backward with the remaining legs like dung beetles. Finally, an insect-inspired robotic (MORF), which is a hexapod robotic platform, demonstrates typical insectlike gaits (gradual wave and quick tripod gaits). In a nutshell, we imagine that this bioinspired robotic zoo demonstrates how the various and interesting talents of residing creatures can function inspiration and rules for creating robotics expertise able to reaching a number of robotic features and fixing complicated motor management issues in techniques with many levels of freedom.

“AngGo: Shared Indoor Smart Mobility Device,” by Yoon Joung Kwak, Haeun Park, Donghun Kang, Byounghern Kim, Jiyeon Lee, and Hui Sung Lee from Ulsan National Institute of Science and Technology (UNIST), in Ulsan, South Korea.

Abstract: AngGo is a hands-free shared indoor good mobility system for public use. AngGo is a private mobility system that’s appropriate for the motion of passengers in big indoor areas similar to conference facilities or airports. The consumer can use each fingers freely whereas driving the AngGo. Unlike current mobility gadgets, the mobility system might be maneuvered utilizing the ft and was designed to be as intuitive as potential. The phrase “AngGo” is pronounced like a Korean phrase that means “sit down and move.” There are 6 ToF distance sensors round AngGo. Half of them are within the entrance half and the opposite half are within the rear half. In the autonomous mode, AngGo avoids obstacles based mostly on the space from every sensor. IR distance sensors are mounted beneath the footrest to measure the extent to which the footrest is moved ahead or backward, and these information are used to manage the rotational velocity of motors. The consumer can management the velocity and the path of AngGo concurrently. The spring within the footrest generates power suggestions, so the consumer can acknowledge the quantity of variation.

“Creative Robotic Pen-Art System,” by Daeun Song and Young Jun Kim from Ewha Womans University in Seoul, South Korea.

Abstract: Since the Renaissance, artists have created artworks utilizing novel methods and machines, deviating from typical strategies. The robotic drawing system is one in every of such artistic makes an attempt that includes not solely the creative nature but additionally scientific issues that should be solved. Robotic drawing issues might be seen as planning the robotic’s drawing path that ultimately results in the artwork type. The robotic pen-art system imposes new challenges, not like robotic portray, requiring the robotic to keep up steady contact with the goal drawing floor. This video showcases an autonomous robotic system that creates pen artwork on an arbitrary canvas floor with out proscribing its measurement or form. Our system converts raster or vector photos into piecewise-continuous paths relying on stylistic decisions, similar to TSP artwork or stroke-based drawing. Our system consists of a number of manipulators with mobility and performs stylistic drawing duties. In order to create a extra intensive pen artwork, the cell manipulator setup finds a minimal variety of discrete configurations for the cell platform to cowl the ample canvas area. The twin manipulator setup can generate multicolor pen artwork utilizing adaptive three-finger grippers with a pen-tool-change mechanism. We display that our system can create visually pleasing and complex pen artwork on varied surfaces.

“I Know What You Want: A ‘Smart Bartender’ System by Interactive Gaze Following,” by Haitao Lin, Zhida Ge, Xiang Li, Yanwei Fu, and Xiangyang Xue from Fudan University, in Shanghai, China.

Abstract: We developed a novel “Smart Bartender” system, which may perceive the intention of customers simply from the attention gaze and make some corresponding actions. Particularly, we imagine {that a} cyber-barman who can not really feel our faces will not be an clever one. We thus purpose at constructing a novel cyber-barman by capturing and analyzing the intention of the shoppers on the fly. Technically, such a system permits the consumer to pick a drink just by watching it. Then the robotic arm mounted with a digital camera will mechanically grasp the goal bottle and pour the liquid into the cup. To obtain this aim, we firstly undertake YOLO to detect candidate drinks. Then, the GazeWeb is utilized to generate potential gaze heart for grounding the goal bottle that has minimal center-to-center distance. Finally, we use object pose estimation and path-planning algorithms to information the robotic arm to know the goal bottle and execute pouring. Our system built-in with the category-level object pose estimation enjoys highly effective efficiency, generalizing to numerous unseen bottles and cups that aren’t used for coaching. We imagine our system wouldn’t solely cut back the intensive human labor in numerous service eventualities but additionally present customers with interactivity and pleasure.

“Towards Aerial Humanoid Robotics: Developing the Jet-Powered Robot iRonCub,” by Daniele Pucci, Gabriele Nava, Fabio Bergonti, Fabio Di Natale, Antonello Paolino, Giuseppe L’erario, Affaf Junaid Ahamad Momin, Hosameldin Awadalla Omer Mohamed, Punith Reddy Vanteddu, and Francesca Bruzzone from the Italian Institute of Technology (IIT), in Genoa, Italy.

Abstract: The present state of robotics expertise lacks a platform that may mix manipulation, aerial locomotion, and bipedal terrestrial locomotion. Therefore, we outline aerial humanoid robotics as the end result of platforms with these three capabilities. To implement aerial humanoid robotics on the humanoid robotic iCub, we conduct analysis in numerous instructions. This contains experimental analysis on jet generators and codesign, which is important to implement aerial humanoid robotics on the actual iCub. These actions purpose to mannequin and establish the jet generators. We additionally examine flight management of flying humanoid robots utilizing Lyapunov-quadratic-programming-based management algorithms to control each the perspective and place of the robotic. These algorithms work independently of the variety of jet generators put in on the robotic and guarantee satisfaction of bodily constraints related to the jet engines. In addition, we analysis computational fluid dynamics for aerodynamics modeling. Since the aerodynamics of a multibody system like a flying humanoid robotic is complicated, we use CFD simulations with Ansys to extract a simplified mannequin for management design, as there’s little area for closed-form expressions of aerodynamic results.

“AMEA Autonomous Electrically Operated One-Axle Mowing Robot,” by Romano Hauser, Matthias Scholer, and Katrin Solveig Lohan from Eastern Switzerland University of Applied Sciences (OST), in St. Gallen, Switzerland, and Heriot-Watt University, in Edinburgh, Scotland.

Abstract: The aim of this analysis venture (Consortium: Altatek GmbH, Eastern Switzerland University of Applied Sciences OST, Faculty of Law University of Zurich) was the event of a multifunctional, autonomous single-axle robotic with an electrical drive. The robotic is custom-made for agricultural functions in mountainous areas with steepest slopes. The intention is to alleviate farmers from arduous and safety-critical work. Furthermore, the robotic is developed as a modular platform that can be utilized for work in forestry, municipal, sports activities fields, and winter/snow functions. Robot options: Core function is the patented heart of gravity management. With a sliding wheel axle of 800 millimeters, hills as much as a steepness of 35 levels (70 p.c) might be simply pushed and a protected operation with out tipping might be ensured. To make the robotic extra sustainable, electrical drives and a 48-volt battery have been outfitted. To navigate in mountainous areas, a number of sensors are used. In distinction to functions on flat areas, the place and gradient of the robotic on the slope must be measured and regarded within the path planning. A sensor system that detects potential obstacles and particularly people or animals which might be within the path of the robotic is presently beneath growth.

“Surf Zone Exploration With Crab-Like Legged Robots,” by Yifeng Gong, John Grezmak, Jianfeng Zhou, Nicole Graf, Zhili Gong, Nathan Carmichael, Airel Foss, Glenna Clifton, and Kathryn A. Daltorio from Case Western Reserve University, in Cleveland, and the University of Portland, in Oregon.

Abstract: Surf zones are difficult for strolling robots if they can’t anchor to the substrate, particularly on the transition between dry sand and waves. Crablike dactyl designs allow robots to realize this anchoring habits whereas nonetheless being light-weight sufficient to stroll on dry sand. Our group has been creating a collection of crablike robots to realize the transition from strolling on underwater surfaces to strolling on dry land. Compared with the default forward-moving gait, we discover that inward-pulling gaits and sideways strolling improve effectivity in granular media. By utilizing gentle dactyls, robots can probe the bottom to categorise substrates, which may also help modify gaits to raised go well with the surroundings and acknowledge hazardous situations. Dactyls may also be used to securely grasp the thing and dig within the substrate for putting in cables, trying to find buried objects, and accumulating sediment samples. To simplify management and actuation, we developed a four-degrees-of-freedom Klann mechanism robotic, which may climb onto an object after which grasp it. In addition, human interfaces will enhance our means to exactly management the robotic for these kind of duties. In explicit, the U.S. authorities has recognized munitions retrieval as an environmental precedence via their Strategic Environmental Research Development Program. Our aim is to assist these efforts with new robots.

“Learning Exploration Strategies to Solve Real-World Marble Runs,” by Alisa Allaire and Christopher G. Atkeson from the Robotics Institute at Carnegie Mellon University, in Pittsburgh.

Abstract: Tasks involving domestically unstable or discontinuous dynamics (similar to bifurcations and collisions) stay difficult in robotics, as a result of small variations within the surroundings can have a major affect on process outcomes. For such duties, studying a strong deterministic coverage is tough. We give attention to structuring exploration with a number of stochastic insurance policies based mostly on a mix of specialists (MoE) coverage illustration that may be effectively tailored. The MoE coverage consists of stochastic subpolicies that permit exploration of a number of distinct areas of the motion area (or methods) and a high- degree choice coverage to information exploration towards essentially the most promising areas. We develop a robotic system to guage our strategy in a real-world bodily problem-solving area. After coaching the MoE coverage in simulation, on-line studying in the actual world demonstrates environment friendly adaptation inside just some dozen makes an attempt, with a minimal sim2real hole. Our outcomes affirm that representing a number of methods promotes environment friendly adaptation in new environments and techniques discovered beneath completely different dynamics can nonetheless present helpful details about the place to search for good methods.

“Flipbot: Learning Continuous Paper-Flipping Via Coarse-to-Fine Exteroceptive-Proprioceptive Exploration,” by Chao Zhao, Chunli Jiang, Junhao Cai, Michael Yu Wang, Hongyu Yu, and Qifeng Chen from Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong, and HKUST-Shenzhen-Hong Kong Collaborative Innovation Research Institute, Futian, Shenzhen.

Abstract: This paper tackles the duty of singulating and greedy paperlike deformable objects. We confer with such duties as paper-flipping. In distinction to manipulating deformable objects that lack compression power (similar to shirts and ropes), minor variations within the bodily properties of the paperlike deformable objects considerably affect the outcomes, making manipulation extremely difficult. Here, we current Flipbot, a novel resolution for flipping paperlike deformable objects. Flipbot permits the robotic to seize object bodily properties by integrating exteroceptive and proprioceptive perceptions which might be indispensable for manipulating deformable objects. Furthermore, by incorporating a proposed coarse-to-fine exploration course of, the system is able to studying the optimum management parameters for efficient paper-flipping via proprioceptive and exteroceptive inputs. We deploy our technique on a real-world robotic with a gentle gripper and study in a self-supervised method. The ensuing coverage demonstrates the effectiveness of Flipbot on paper-flipping duties with varied settings past the attain of prior research, together with however not restricted to flipping pages all through a e book and emptying paper sheets in a field. The code is offered right here:
https://robotll.github.io/Flipbot/

“Croche-Matic: A Robot for Crocheting 3D Cylindrical Geometry,” by Gabriella Perry, Jose Luis Garcia del Castillo y Lopez, and Nathan Melenbrink from Harvard University, in Cambridge, Mass.

Abstract: Crochet is a textile craft that has resisted mechanization and industrialization aside from a choose variety of one-off crochet machines. These machines are solely able to producing a restricted subset of frequent crochet stitches. Crochet machines aren’t used within the textile trade, but mass-produced crochet objects and garments offered in shops like Target and Zara are nearly actually the merchandise of crochet sweatshops. The recognition of crochet and the existence of crochet merchandise in main chain shops exhibits that there’s each a transparent demand for this craft in addition to a necessity for it to be produced in a extra moral approach. In this paper, we current Croche-Matic, a radial crochet machine for producing three-dimensional cylindrical geometry. The Croche-Matic is designed based mostly on Magic Ring approach, a way for hand-crocheting 3D cylindrical objects. The machine consists of 9 mechanical axes that work in sequence to finish several types of crochet stitches, and features a sensor part for measuring and regulating yarn pressure throughout the mechanical system. Croche-Matic can full the 4 foremost stitches utilized in Magic Ring approach. It has successful charge of fifty.7 p.c with single crochet stitches, and has demonstrated a capability to create three-dimensional objects.

“SOPHIE: SOft and Flexible Aerial Vehicle for PHysical Interaction with the Environment,” by F. Ruiz , B. C. Arrue, and A. Ollero from GRVC Robotics Lab of Seville, Spain.

Abstract: This letter presents the primary design of a gentle and light-weight UAV, solely 3D-printed in versatile filament. The drone’s versatile arms are outfitted with a tendon-actuated bending system, which is used for functions that require bodily interplay with the surroundings. The flexibility of the UAV might be managed in the course of the additive manufacturing course of by adjusting the infill charge ρTPU distribution. However, the rise inflexibility implies difficulties in controlling the UAV, in addition to structural, aerodynamic, and aeroelastic results. This article supplies perception into the dynamics of the system and validates the flyability of the car for densities as little as 6 p.c. Within this vary, quasi-static arm deformations might be thought of; thus the autopilot is fed again via a static arm deflection mannequin. At decrease densities, robust nonlinear elastic dynamics seem, which interprets to complicated modeling, and it’s advised to modify to data-based approaches. Moreover, this work demonstrates the power of the gentle UAV to carry out full-body perching, particularly touchdown and stabilizing on pipelines and irregular surfaces with out the necessity for an auxiliary system.

“Reconfigurable Drone System for Transportation of Parcels with Variable Mass and Size,” by Fabrizio Schiano, Przemyslaw Mariusz Kornatowski, Leonardo Cencetti, and Dario Floreano from École Polytechnique Fédérale de Lausanne (EPFL), in Switzerland, and Leonardo S.p.A., Leonardo Labs, in Rome.

Abstract: Cargo drones are designed to hold payloads with predefined form, measurement, and/or mass. This lack of flexibility requires a fleet of numerous drones tailor-made to particular cargo dimensions. Here we suggest a brand new reconfigurable drone based mostly on a modular design that adapts to completely different cargo shapes, sizes, and mass. We additionally suggest a way for the automated technology of drone configurations and appropriate parameters for the flight controller. The parcel turns into the drone’s physique to which a number of particular person propulsion modules are hooked up. We display the usage of the reconfigurable {hardware} and the accompanying software program by transporting parcels of various mass and sizes requiring varied numbers and propulsion modules’ positioning. The experiments are performed indoors (with a motion-capture system) and outside (with an RTK-GNSS sensor). The proposed design represents a less expensive and extra versatile various to the options involving a number of drones for parcel transportation.