Learning to swim in potential flow

Jiao, Yusheng; Ling, Feng; Heydari, Sina; Heess, Nicolas; Merel, Josh; Kanso, Eva

doi:10.1103/physrevfluids.6.050505

Cited by 34 publications

(22 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…2016; Verma, Novati & Koumoutsakos 2018) or in the potential flow (Jiao et al. 2021). In particular, recent pioneering experiments (Muiños-Landin et al.…”

Section: Problem Set-up Assumptions and Methodsmentioning

confidence: 99%

“…It is worth noting that RL algorithms have been used recently in similar swimming-involved scenarios, e.g. to optimize the swimming gaits or navigation routes of microswimmers at low Reynolds number (Colabrese et al 2017;Schneider & Stark 2019;Mirzakhanloo et al 2020;Tsang et al 2020;Muiños-Landin et al 2021;Nasiri & Liebchen 2022;Qiu et al 2022) and in turbulent flows (Alageshan et al 2020;Qiu et al 2020), or macroscopic swimmers such as fish in viscous flows (Gazzola et al 2016;Verma, Novati & Koumoutsakos 2018) or in the potential flow (Jiao et al 2021). In particular, recent pioneering experiments (Muiños-Landin et al 2021) have demonstrated using RL for real-time navigation of micron-sized thermophoretic particles, opening a new horizon for developing swimming microrobots endowed with artificial intelligence.…”

Section: Reinforcement Learning For a Point Or Finite-size Preymentioning

confidence: 99%

See 1 more Smart Citation

Optimizing low-Reynolds-number predation via optimal control and reinforcement learning

2022

View full text Add to dashboard Cite

We seek the best stroke sequences of a finite-size swimming predator chasing a non-motile point or finite-size prey at low Reynolds number. We use optimal control to seek the globally optimal solutions for the former and reinforcement learning (RL) for general situations. The predator is represented by a squirmer model that can translate forward and laterally, rotate and generate a stresslet flow. We identify the predator's best squirming sequences to achieve the time-optimal (TO) and efficiency-optimal (EO) predation. For a point prey, the TO squirmer executing translational motions favours a two-fold $L$ -shaped trajectory that enables it to exploit the disturbance flow for accelerated predation; using a stresslet mode expedites significantly the EO predation, allowing the predator to catch the prey faster yet with lower energy consumption and higher predatory efficiency; the predator can harness its stresslet disturbance flow to suck the prey towards itself; compared to a translating predator, its compeer combining translation and rotation is less time-efficient, and the latter occasionally achieves the TO predation via retreating in order to advance. We also adopt RL to reproduce the globally optimal predatory strategy of chasing a point prey, qualitatively capturing the crucial two-fold attribute of a TO path. Using a numerically emulated RL environment, we explore the dependence of the optimal predatory path on the size of prey. Our results might provide useful information that help in the design of synthetic microswimmers such as in vivo medical microrobots capable of capturing and approaching objects in viscous flows.

show abstract

“…2016; Verma, Novati & Koumoutsakos 2018) or in the potential flow (Jiao et al. 2021). In particular, recent pioneering experiments (Muiños-Landin et al.…”

Section: Problem Set-up Assumptions and Methodsmentioning

confidence: 99%

Section: Reinforcement Learning For a Point Or Finite-size Preymentioning

confidence: 99%

Optimizing low-Reynolds-number predation via optimal control and reinforcement learning

2022

View full text Add to dashboard Cite

show abstract

“…Rabault studied the application of DRL in active flow control and shape optimization and Xu utilized DRL to find out the optimal control strategy for rotating cylinders [257,258]. Jiao trained fish to learn how to swim in potential flow based on RL [259]. Still, the research on fish locomotion mechanism based on ML is not near enough.…”

Section: Machine Learningmentioning

confidence: 99%

Research Development on Fish Swimming

Jiang

2022

Chin. J. Mech. Eng.

View full text Add to dashboard Cite

Fishes have learned how to achieve outstanding swimming performance through the evolution of hundreds of millions of years, which can provide bio-inspiration for robotic fish design. The premise of designing an excellent robotic fish include fully understanding of fish locomotion mechanism and grasp of the advanced control strategy in robot domain. In this paper, the research development on fish swimming is presented, aiming to offer a reference for the later research. First, the research methods including experimental methods and simulation methods are detailed. Then the current research directions including fish locomotion mechanism, structure and function research and bionic robotic fish are outlined. Fish locomotion mechanism is discussed from three views: macroscopic view to find a unified principle, microscopic view to include muscle activity and intermediate view to study the behaviors of single fish and fish school. Structure and function research is mainly concentrated from three aspects: fin research, lateral line system and body stiffness. Bionic robotic fish research focuses on actuation, materials and motion control. The paper concludes with the future trend that curvature control, machine learning and multiple robotic fish system will play a more important role in this field. Overall, the intensive and comprehensive research on fish swimming will decrease the gap between robotic fish and real fish and contribute to the broad application prospect of robotic fish.

show abstract

“…6, RL does not rely on prior knowledge of the dynamics but allows the squirmer as the predating agent to learn the dynamics, adapt and optimize its chasing strategy (or policy in the language of RL) via continuously interacting with the environment. It is worth-noting that RL algorithms have been recently used in similar swimming-involved scenarios, e.g., to optimize the swimming gaits or navigation routes of micro-swimmers at low Reynolds num-ber [8,37,39,40,46,49,57] and in turbulent flows [2,45], macroscopic swimmers such as fish in viscous flows [15,58] or in the potential flow [20]. Particularly, the recent pioneering experiments [39] have demonstrated using RL for real-time navigation of micron-sized thermophoretic particles, opening a new horizon for developing swimming micro-robots endowed with artificial intelligence.…”

Section: B Optimal Control For a Point Preymentioning

confidence: 99%

Optimising low-Reynolds-number predation via optimal control and reinforcement learning

Zhu,

Fang,

Zhu

2022

Preprint

View full text Add to dashboard Cite

We seek the best stroke sequences of a finite-size swimming predator chasing a non-motile point or finite-size prey at low Reynolds number. We use optimal control to seek the globally-optimal solutions for the former and RL for general situations. The predator is represented by a squirmer model that can translate forward and laterally, rotate and generate a stresslet flow. We identify the predator's best squirming sequences to achieve the time-optimal (TO) and efficiency-optimal (EO) predation. For a point prey, the TO squirmer executing translational motions favours a two-fold L-shaped trajectory that enables it to exploit the disturbance flow for accelerated predation; using a stresslet mode significantly expedites the EO predation, allowing the predator to catch the prey faster yet with lower energy consumption and higher predatory efficiency; the predator can harness its stresslet disturbance flow to suck the prey towards itself; compared to a translating predator, its compeer combining translation and rotation is less time-efficient, and the latter occasionally achieves the TO predation via retreating in order to advance. We also adopt RL to reproduce the globally-optimal predatory strategy of chasing a point prey, qualitatively capturing the crucial two-fold attribute of TO path. Using a numerically emulated RL environment, we explore the dependence of the optimal predatory path on the size of prey. Our results might provide useful information that help design synthetic microswimmers such as in vivo medical micro-robots capable of capturing and approaching objects in viscous flows.

show abstract

Learning to swim in potential flow

Cited by 34 publications

References 47 publications

Optimizing low-Reynolds-number predation via optimal control and reinforcement learning

Optimizing low-Reynolds-number predation via optimal control and reinforcement learning

Research Development on Fish Swimming

Optimising low-Reynolds-number predation via optimal control and reinforcement learning

Contact Info

Product

Resources

About