The fish-like robot is one of the typical underwater robots, which has the advantage of high maneuverability with low noise due to its bioinspired structure and biomimetic locomotion. However, it is challenging to efficiently design motion controllers for such robots to achieve satisfactory performance on specific control tasks in the real underwater environment, since the complex fluid-structure interaction exists during their swimming and exact dynamic models are absent. In this article, we propose a learning framework, incorporating a simulation system and a training methodology, to autonomously and fast train in simulation to create control policies that are capable of directly applying to a type of physical fish-like robots to perform motion control tasks. First, we construct a simulation system combining a data-driven environment and a computational fluid dynamics (CFD)-based environment, thus well balancing the simulation accuracy and the calculation speed. Second, we design a training methodology to train deep reinforcement learning (DRL)-based policies for the robot in our constructed simulation system to perform a specific control task. Then, we use two typical motion control tasks to verify our proposed framework. One is the path-following control task, which is a one-objective problem with dense rewards, while the other is the pose control task which is a two-objective problem with sparse rewards. For each task, the DRL-based control policy trained by our learning framework is directly deployed on the physical fish-like robot to perform the task in the real world. Experimental results show that the policies trained in simulation still work well in the real world, and perform even better in terms of control accuracy and stability compared with the traditional control methods, thus demonstrating the effectiveness of our learning framework.

Controlling biomimetic underwater robots in unknown flow fields remains a challenge due to the strong nonlinearity of the fluid. This article investigates the attitude holding task of a robotic fish swimming in reality. Such a typical sensing-based control task requires the fish to keep a desired angle of attack in an unknown and even varied incoming flow. To this end, we propose a learning-based approach by using a deep neural network directly maps the raw data of sensors equipped on the robot to the continuous control signals in an end-to-end manner. First, based on experimental data of the physical robot, we construct a data-driven simulation environment including three modules of dynamic, sensor, and control. The dynamic and sensor modules are established to model the dynamics of the fish and to generate its sensors’ data, based on which a deep reinforcement learning (DRL) algorithm in the control module is trained to get a control policy. Then, we directly deploy the trained policy to a physical robotic fish for attitude holding task. Experimental results demonstrate the robustness and effectiveness of the DRL policy and, thus, verify the success of our approach to achieving sim-to-real transfer.

Possessing the attributes of high adaptability and low cost, soft robotic individuals can further coordinate and form into a swarming system, enhancing the performances as well as functions in practical applications. However, the formation control of soft robotic swarm remains challenging mainly due to the limitation in relatively low precision and slow response of the soft actuators. In this work, a soft robotic fish swarm system with global vision positioning was studied. The soft robotic fish used in the project is driven by a hybrid power-control system, in which the soft dielectric elastomers and the rigid electrical servo provide forward propulsion and controllable steering function, respectively. Results show that soft robotic fish swarm can quickly shift their formations, mimicking three typical swarming behaviors of natural creatures: highly parallel group, encircling, and torus. The system design and controlling principles of the soft robotic fish swarm may guide the future research of soft robots and robotic swarms, specifically for underwater applications.

Attention-Deficit/Hiperactivity Disorder (ADHD) is a prevalent disorder, but its neuroanatomical circuitry is still relatively understudied, especially in the adult population. The few morphometric magnetic resonance imaging (MRI) and diffusion tensor imaging (DTI) studies available to date have found heterogeneous results. This may be at least partly attributable to some well-known technical limitations of the conventional voxel-based methods usually employed to analyze such neuroimaging data. Moreover, there is a great paucity of imaging studies of adult ADHD to date that have excluded patients with history of use of stimulant medication. A newly validated method named optimally-discriminative voxel-based analysis (ODVBA) was applied to multimodal (structural and DTI) MRI data acquired from 22 treatment-naïve ADHD adults and 19 age- and gender-matched healthy controls (HC). Regarding DTI data, we found higher fractional anisotropy in ADHD relative to HC encompassing the white matter (WM) of the bilateral superior frontal gyrus, right middle frontal left gyrus, left postcentral gyrus, bilateral cingulate gyrus, bilateral middle temporal gyrus and right superior temporal gyrus; reductions in trace (a measure of diffusivity) in ADHD relative to HC were also found in fronto-striatal-parieto-occipital circuits, including the right superior frontal gyrus and bilateral middle frontal gyrus, right precentral gyrus, left middle occipital gyrus and bilateral cingulate gyrus, as well as the left body and right splenium of the corpus callosum, right superior corona radiata, and right superior longitudinal and fronto-occipital fasciculi. Volumetric abnormalities in ADHD subjects were found only at a trend level of significance, including reduced gray matter (GM) in the right angular gyrus, and increased GM in the right supplementary motor area and superior frontal gyrus. Our results suggest that adult ADHD is associated with neuroanatomical abnormalities mainly affecting the WM microstructure in fronto-parieto-temporal circuits that have been implicated in cognitive, emotional and visuomotor processes.

In nature, the lateral line system (LLS) is a critical sensor organ of fish for rheotaxis in complex environments. Inspired by the LLS, numbers of artificial lateral line systems (ALLSs) have been designed to the fish-like robots for flow field perception, assisting the robots to be stable in the face of flow disturbances. However, almost all pressure sensor based ALLSs are facing the challenge of the low signal to noise ratio (SNR), resulting in inaccurate perception information. For solving this problem, this paper describes a dual-sensor fusion method by integrating the ALLSs with the inertial measurement unit (IMU), and shows the excellent performance by a higher precision and lower latency attitude holding of robotic fish. First, low-pass filtering is performed on ALLS data with low-SNR. Second, the ALLS data is mapped to the angle of attack based on an artificial neural network. Finally, a fusion perception method is established based on the time correlation between ALLS and IMU. To demonstrate the efficacy of our proposed method, we compare the result of attitude holding by three methods (dual-sensor fusion method, IMU based method, and ALLS based method). Furthermore, dual-sensor fusion method is tested at varied flow velocities and varied desired angles of attack, indicating that the algorithm can enable the robotic fish to perform dynamic movements in the incoming flow. This work provides a method for the attitude control of autonomous underwater vehicles (AUVs) by fusing the sensory data of ALLS and IMU, which is also applicable to other flow sensors and IMU.