Real-Time Learning of Efficient Lift Generation on a Dynamically Scaled Flapping Wing Using Policy Search

Bayiz, Yagiz E.; Chen, Long; Hsu, Shih-Jung; Liu, Pan; Aguiles, Aaron N.; Cheng, Bo

doi:10.1109/icra.2018.8460781

Cited by 9 publications

(12 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The model was trained using data acquired from a dynamically scaled robotic wing [21,22,53] (figure 2 b ). In this work, we aimed to develop a general predictive model valid for a broad range of different wing motion kinematics while providing a novel and fundamental perspective towards flapping flight—characterized by fast, reciprocal, and 3DoF motion of the aerodynamic surface.…”

Section: Prssm As a Predictive Model Of Flapping Wing Aerodynamicsmentioning

confidence: 99%

State-space aerodynamic model reveals high force control authority and predictability in flapping flight

Bayiz

Cheng

2021

J. R. Soc. Interface.

Self Cite

View full text Add to dashboard Cite

Flying animals resort to fast, large-degree-of-freedom motion of flapping wings, a key feature that distinguishes them from rotary or fixed-winged robotic fliers with limited motion of aerodynamic surfaces. However, flapping-wing aerodynamics are characterized by highly unsteady and three-dimensional flows difficult to model or control, and accurate aerodynamic force predictions often rely on expensive computational or experimental methods. Here, we developed a computationally efficient and data-driven state-space model to dynamically map wing kinematics to aerodynamic forces/moments. This model was trained and tested with a total of 548 different flapping-wing motions and surpassed the accuracy and generality of the existing quasi-steady models. This model used 12 states to capture the unsteady and nonlinear fluid effects pertinent to force generation without explicit information of fluid flows. We also provided a comprehensive assessment of the control authority of key wing kinematic variables and found that instantaneous aerodynamic forces/moments were largely predictable by the wing motion history within a half-stroke cycle. Furthermore, the angle of attack, normal acceleration and pitching motion had the strongest effects on the aerodynamic force/moment generation. Our results show that flapping flight inherently offers high force control authority and predictability, which can be key to developing agile and stable aerial fliers.

show abstract

Section: Prssm As a Predictive Model Of Flapping Wing Aerodynamicsmentioning

confidence: 99%

State-space aerodynamic model reveals high force control authority and predictability in flapping flight

Bayiz

Cheng

2021

J. R. Soc. Interface.

Self Cite

View full text Add to dashboard Cite

show abstract

“…the kinematic parameters in the fin motion trajectories that are need to be optimised. The policy gradient methods commonly maximise the expected return

J^{π}

by using the policy gradient ascent, with a policy update:

θ false\leftarrow θ + γ \nabla_{π} J^{π}

where

γ

is the learning rate,

\nabla_{π} J^{π}

is the policy gradient and

θ

is the policy parameter vector [32]. In episode‐based algorithms, policy gradient is estimated using the total cumulative reward for several rollouts (trials) that share the same policy.…”

Section: Methodsmentioning

confidence: 99%

“…The PEPG algorithm, on the other hand, learns distributions of policy parameters rather than themselves, shifting the exploration from action space to parameter space, reducing the variance. In turn, the reliability of the algorithm increases significantly [32], while quality and speed of convergence is improved. Particularly, in PEPG,

μ

and

σ

(Gaussian distribution assumed) are introduced, which represent mean and standard deviation of policy parameter vector

θ

, and

θ \sim N false(μ, I σ^{2} false)

, and these distribution parameters are updated using the policy gradients.…”

Section: Methodsmentioning

confidence: 99%

“…fin flapping cycle(s) with a time length T. Therefore, in this work, an episode-based learning algorithm, named parameter exploring policy gradient (PEPG) is used [31]. This algorithm has been successfully applied to several robot-learning problems including an analogous periodic locomotion in fluids problem [32]. In that work, hovering efficiency of a flapping wing at low Re is optimised.…”

Section: Policy Gradient Algorithmmentioning

confidence: 99%

“…where γ is the learning rate, ∇ π J π is the policy gradient and θ is the policy parameter vector [32]. In episode-based algorithms, policy gradient is estimated using the total cumulative reward for several rollouts (trials) that share the same policy.…”

Section: Policy Gradient Algorithmmentioning

confidence: 99%

See 2 more Smart Citations

Efficient thrust generation in robotic fish caudal fins using policy search

Shan

Bayiz

Cheng

2019

IET cyber-systems robotics

Self Cite

View full text Add to dashboard Cite

Hovering efficiency comparison of rotary and flapping flight for rigid rectangular wings via dimensionless multi-objective optimization

et al. 2018

View full text Add to dashboard Cite

In this work, a multi-objective optimization framework is developed for optimizing low Reynolds number ([Formula: see text]) hovering flight. This framework is then applied to compare the efficiency of rigid revolving and flapping wings with rectangular shape under varying [Formula: see text] and Rossby number ([Formula: see text], or aspect ratio). The proposed framework is capable of generating sets of optimal solutions and Pareto fronts for maximizing the lift coefficient and minimizing the power coefficient in dimensionless space, explicitly revealing the trade-off between lift generation and power consumption. The results indicate that revolving wings are more efficient when the required average lift coefficient [Formula: see text] is low (<1 for [Formula: see text] and <1.6 for [Formula: see text]), while flapping wings are more efficient in achieving higher [Formula: see text]. With the dimensionless power loading as the single-objective performance measure to be maximized, rotary flight is more efficient than flapping wings for [Formula: see text] regardless of the amount of energy storage assumed in the flapping wing actuation mechanism, while flapping flight is more efficient for [Formula: see text]. It is observed that wings with low [Formula: see text] perform better when higher [Formula: see text] is needed, whereas higher [Formula: see text] cases are more efficient at [Formula: see text] regions. However, for the selected geometry and [Formula: see text], the efficiency is weakly dependent on [Formula: see text] when the dimensionless power loading is maximized.

show abstract

Real-Time Learning of Efficient Lift Generation on a Dynamically Scaled Flapping Wing Using Policy Search

Cited by 9 publications

References 28 publications

State-space aerodynamic model reveals high force control authority and predictability in flapping flight

State-space aerodynamic model reveals high force control authority and predictability in flapping flight

Efficient thrust generation in robotic fish caudal fins using policy search

Hovering efficiency comparison of rotary and flapping flight for rigid rectangular wings via dimensionless multi-objective optimization

Contact Info

Product

Resources

About