Control and Morphology Optimization of Passive Asymmetric Structures for Robotic Swimming

Obayashi, Nana; Vicari, Andrea; Junge, Kai; Shakir, Kamran; Hughes, Josie

doi:10.1109/lra.2023.3240334

Cited by 6 publications

(4 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An LSTM utilizes past states to make predictions [ 30 ]. This is a neural network architecture that is increasingly being used for tactile sensor classification and detection [ 31 , 32 ]. To train the LSTM network with the acceleration data, the collected dataset was split in a ratio of 70:30, meaning that 70% of the available time series were used for training, and the rest were used for validation.…”

Section: Methodsmentioning

confidence: 99%

Exploration of the Design of Spiderweb-Inspired Structures for Vibration-Driven Sensing

2023

Self Cite

View full text Add to dashboard Cite

In the quest to develop large-area soft sensors, we can look to nature for many examples. Spiderwebs show many fascinating properties that we can seek to understand and replicate in order to develop large-area, soft, and deformable sensing structures. Spiders’ webs are used not only to capture prey, but also to localize their prey through the vibrations that they feel through their legs. Inspired by spiderwebs, we developed a large-area tactile sensor for localizing contact points through vibration sensing. We hypothesize that the structure of a web can be leveraged to amplify, filter, or otherwise morphologically tune vibrations to improve sensing capabilities. To explore this design space, we created a means of computationally designing and 3D printing web structures. By using vibration sensors mounted on the edges of webs to simulate a spider monitoring vibrations, we show how varying the structural properties affects the localization performance when using vibration sensors and long short-term memory (LSTM)-based neural network classifiers. We seek to explain the classification performance seen in different webs by considering various metrics of information content for different webs and, hence, provide insight into how bio-inspired spiderwebs can be used to assist large-area sensing structures.

show abstract

Section: Methodsmentioning

confidence: 99%

Exploration of the Design of Spiderweb-Inspired Structures for Vibration-Driven Sensing

2023

Self Cite

View full text Add to dashboard Cite

show abstract

“…Bayesian optimization is used to perform iterative search of the design space, with each wing geometry tested five times by the robotic setup. Bayesian optimization is useful in trying to capture the stochastic and probabilistic nature of this design problem and allows for a sequential decision making, iteratively sampling new geometries [36][37][38][39][40] . At each iteration of the optimization algorithm, the GMMs for all behaviors are updated.…”

Section: Gmm Based Modelingmentioning

confidence: 99%

Robotic automation and unsupervised cluster assisted modeling for solving the forward and reverse design problem of paper airplanes

Obayashi¹,

Junge²,

Ilić³

et al. 2023

Sci Rep

Self Cite

View full text Add to dashboard Cite

Although often regarded a childhood toy, the design of paper airplanes is subtly complex. The design space and mapping from geometry to distance flown is highly nonlinear and probabilistic where a single airplane design exhibits a multitude of trajectory forms and flight distances. This makes optimization and understanding of their behavior challenging for humans. By understanding the behavior of paper airplanes and predicting flight behavior, there is a potential to improve the design of aerial vehicles that operate at low Reynolds numbers. By developing a robotic system that can fabricate, test, analyze, and model the flight behavior in an unsupervised fashion, a wide design space can be reliably characterized. We find there are discrete behavioral groups that result in different trajectories: nose dive, glide, and recovery glide. Informed by this characterization we propose a method of using Gaussian mixture models to extract the clusters of the design space that map to these different behaviors. This allows us to solve both the forward and reverse design problem for paper airplanes, and also to perform efficient optimization of the geometry for a given target flight distance.

show abstract

“…This approach provides good compensation for the low modeling accuracy caused by the complex coupling between flexible structures and fluids. Data-driven approaches to improve the simulation accuracy based on data show significant potential [ 27 ]. The main problem with this approach currently is the high time cost of the experimental tests and calculations required for data-driven methods.…”

Section: Introductionmentioning

confidence: 99%

Design and Reality-Based Modeling Optimization of a Flexible Passive Joint Paddle for Swimming Robots

Hu,

Xu,

Chen

et al. 2024

Biomimetics

View full text Add to dashboard Cite

Rowing motion with paired propellers is an essential actuation mechanism for swimming robots. Previous work in this field has typically employed flexible propellers to generate a net thrust or torque by using changes in the compliance values of flexible structures under the influence of a fluid. The low stiffness values of the flexible structures restrict the upper limit of the oscillation frequency and amplitude, resulting in slow swimming speeds. Furthermore, complex coupling between the fluid and the propeller reduce the accuracy of flexible propeller simulations. A design of a flexible passive joint paddle was proposed in this study, and a dynamics model and simulation of the paddle were experimentally verified. In order to optimize the straight swimming speed, a data-driven model was proposed to improve the simulation accuracy. The effects of the joint number and controller parameters on the robot’s straight swimming speed were comprehensively investigated. The multi-joint paddle exhibited significantly improved thrust over the single-joint paddle in a symmetric driving mode. The data-driven model reduced the total error of the simulated data of the propulsive force in the range of control parameters to 0.51%. Swimming speed increased by 3.3 times compared to baseline. These findings demonstrate the utility of the proposed dynamics and data-driven models in the multi-objective design of swimming robots.

show abstract

Control and Morphology Optimization of Passive Asymmetric Structures for Robotic Swimming

Cited by 6 publications

References 16 publications

Exploration of the Design of Spiderweb-Inspired Structures for Vibration-Driven Sensing

Exploration of the Design of Spiderweb-Inspired Structures for Vibration-Driven Sensing

Robotic automation and unsupervised cluster assisted modeling for solving the forward and reverse design problem of paper airplanes

Design and Reality-Based Modeling Optimization of a Flexible Passive Joint Paddle for Swimming Robots

Contact Info

Product

Resources

About