Cyborg and bionic systems (Washington, D.C.)最新文献

In this paper, the hydrodynamic modeling and parameter identification of the RobDact, a bionic underwater vehicle inspired by Dactylopteridae, are carried out based on computational fluid dynamics (CFD) and force measurement experiment. Firstly, the paper briefly describes the RobDact, then establishes the kinematics model and rigid body dynamics model of the RobDact according to the hydrodynamic force and moment equations. Through CFD simulations, the hydrodynamic force of the RobDact at different speeds is obtained, and then, the hydrodynamic model parameters are identified. Furthermore, the measurement platform is developed to obtain the relationship between the thrust generated by the RobDact and the input fluctuation parameters. Finally, by combining the rigid body dynamics model and the fin thrust mapping model, the hydrodynamic model of the RobDact at different motion states is constructed.

Untethered microrobots have attracted extensive attention due to their potential for biomedical applications and micromanipulation at the small scale. Soft microrobots are of great research importance because of their highly deformable ability to achieve not only multiple locomotion mechanisms but also minimal invasion to the environment. However, the existing microrobots are still limited in their ability to locomote and cross obstacles in unstructured environments compared to conventional legged robots. Nature provides much inspiration for developing miniature robots. Here, we propose a bionic quadruped soft thin-film microrobot with a nonmagnetic soft body and 4 magnetic flexible legs. The quadruped soft microrobot can achieve multiple controllable locomotion modes in the external magnetic field. The experiment demonstrated the robot's excellent obstacle-crossing ability by walking on the surface with steps and moving in the bottom of a stomach model with gullies. In particular, by controlling the conical angle of the external conical magnetic field, microbeads gripping, transportation, and release of the microrobot were demonstrated. In the future, the quadruped microrobot with excellent obstacle-crossing and gripping capabilities will be relevant for biomedical applications and micromanipulation.

Human action representation is derived from the description of human shape and motion. The traditional unsupervised 3-dimensional (3D) human action representation learning method uses a recurrent neural network (RNN)-based autoencoder to reconstruct the input pose sequence and then takes the midlevel feature of the autoencoder as representation. Although RNN can implicitly learn a certain amount of motion information, the extracted representation mainly describes the human shape and is insufficient to describe motion information. Therefore, we first present a handcrafted motion feature called pose flow to guide the reconstruction of the autoencoder, whose midlevel feature is expected to describe motion information. The performance is limited as we observe that actions can be distinctive in either motion direction or motion norm. For example, we can distinguish "sitting down" and "standing up" from motion direction yet distinguish "running" and "jogging" from motion norm. In these cases, it is difficult to learn distinctive features from pose flow where direction and norm are mixed. To this end, we present an explicit pose decoupled flow network (PDF-E) to learn from direction and norm in a multi-task learning framework, where 1 encoder is used to generate representation and 2 decoders are used to generating direction and norm, respectively. Further, we use reconstructing the input pose sequence as an additional constraint and present a generalized PDF network (PDF-G) to learn both motion and shape information, which achieves state-of-the-art performances on large-scale and challenging 3D action recognition datasets including the NTU RGB+D 60 dataset and NTU RGB+D 120 dataset.