Design and control of soft biomimetic pangasius fish robot using fin ray effect and reinforcement learning

Abstract

Soft robots provide a pathway to accurately mimic biological creatures and be integrated into their environment with minimal invasion or disruption to their ecosystem. These robots made from soft deforming materials possess structural properties and behaviors similar to the bodies and organs of living creatures. However, they are difficult to develop in terms of integrated actuation and sensing, accurate modeling, and precise control. This article presents a soft-rigid hybrid robotic fish inspired by the Pangasius fish. The robot employs a flexible fin ray tail structure driven by a servo motor, to act as the soft body of the robot and provide the undulatory motion to the caudal fin of the fish. To address the modeling and control challenges, reinforcement learning (RL) is proposed as a model-free control strategy for the robot fish to swim and reach a specified target goal. By training and investigating the RL through experiments on real hardware, we illustrate the capability of the fish to learn and achieve the required task.

A Survey on Reinforcement Learning Methods in Bionic Underwater Robots

“…Taking into account both attitude control and position control, 2D or 3D path tracking is a common motion control task for bionic underwater robots [ 98 , 105 , 106 , 107 , 108 , 109 , 110 , 111 , 128 , 129 ] that enables the motion capability of bionic robots in underwater environments. The authors of [ 111 ] deployed the DDQN algorithm to the path tracking control of a hybrid-driven robotic fish, and quantitatively compared the control performance of RL with PID and SMC.…”

“…To our knowledge, this study is the first to adopt this technique. Meanwhile, the study by [ 105 ] verified three RL algorithms of PPO, A2C, and DQN in the motion control environment of a soft bionic Pangasius fish robot, and ultimately, the PPO agent performed better. Similarly, a cooperative structured control based on evolutionary strategy and DDPG is proposed for the 3D trajectory tracking control of bionic robotic fish, saving 23.97%, 22.13%, and 38.72% energy compared with SMC, ADRC, and PID, respectively.…”

“…For complex tasks that consider both position and attitude control, reinforcement learning exhibits strong performance, and can effectively address multi-objective control problems. In particular, motion control based on RL has been widely researched and validated in the field of soft robotics [ 81 , 82 , 105 , 107 ]. For instance, the multi-objective control problem of the soft robotic fish with a bundled SCP actuator, which includes heading control and path tracking, was solved by a DDPG controller and a linear–quadratic-regulator-based multi-objective reward mechanism [ 107 ].…”

Continuum Robots and Magnetic Soft Robots: From Models to Interdisciplinary Challenges for Medical Applications

“…Various machine-learning models have been successfully employed in various application scenarios of continuum robots. These models include neural networks [ 142 , 143 , 144 ], reinforcement learning [ 145 ], support vector machines [ 146 ], and a myriad of combined strategies [ 147 ]. They have demonstrated exceptional performance in trajectory prediction, action recognition, and fault detection.…”

Oceanic Challenges to Technological Solutions: A Review of Autonomous Underwater Vehicle Path Technologies in Biomimicry, Control, Navigation, and Sensing

“…-Undulating motion of the fins may create more turbulence and drag in the water. [48] HABH jellyfish -Hydraulic actuators selected for higher forces and speeds.…”