Variational Information Bottleneck Regularized Deep Reinforcement Learning for Efficient Robotic Skill Adaptation.

Deep Reinforcement Learning (DRL) algorithms have been widely studied for sequential decision-making problems, and substantial progress has been achieved, especially in autonomous robotic skill learning. However, it is always difficult to deploy DRL methods in practical safety-critical robot systems, since the training and deployment environment gap always exists, and this issue would become increasingly crucial due to the ever-changing environment. Aiming at efficiently robotic skill transferring in a dynamic environment, we present a meta-reinforcement learning algorithm based on a variational information bottleneck. More specifically, during the meta-training stage, the variational information bottleneck first has been applied to infer the complete basic tasks for the whole task space, then the maximum entropy regularized reinforcement learning framework has been used to learn the basic skills consistent with that of basic tasks. Once the training stage is completed, all of the tasks in the task space can be obtained by a nonlinear combination of the basic tasks, thus, the according skills to accomplish the tasks can also be obtained by some way of a combination of the basic skills. Empirical results on several highly nonlinear, high-dimensional robotic locomotion tasks show that the proposed variational information bottleneck regularized deep reinforcement learning algorithm can improve sample efficiency by 200-5000 times on new tasks. Furthermore, the proposed algorithm achieves substantial asymptotic performance improvement. The results indicate that the proposed meta-reinforcement learning framework makes a significant step forward to deploy the DRL-based algorithm to practical robot systems.

Research on obstacle avoidance algorithm for unmanned ground vehicle based on multi-sensor information fusion.

With the wide application of unmanned ground vehicles (UGV) in a complex environment, the research on the obstacle avoidance system has gradually become an important research part in the field of the UGV system. Aiming at the complex working environment, a sensor detection system mounted on UGV is designed and the kinematic estimation model of UGV is studied. In order to meet the obstacle avoidance requirements of UGVs in a complex environment, a fuzzy neural network obstacle avoidance algorithm based on multi-sensor information fusion is designed in this paper. MATLAB is used to simulate the obstacle avoidance algorithm. By comparing and analyzing the simulation path of UGV's obstacle avoidance motion under the navigation control of fuzzy controller and fuzzy neural network algorithm, the superiority of the proposed fuzzy neural network algorithm was verified. Finally, the superiority and reliability of the obstacle avoidance algorithm are verified through the obstacle avoidance experiment on the UGV experimental platform.

Bending Response of 3D-Printed Titanium Alloy Sandwich Panels with Corrugated Channel Cores.

Ultralight sandwich constructions with corrugated channel cores (i.e., periodic fluid-through wavy passages) are envisioned to possess multifunctional attributes: simultaneous load-carrying and heat dissipation via active cooling. Titanium alloy (Ti-6Al-4V) corrugated-channel-cored sandwich panels (3CSPs) with thin face sheets and core webs were fabricated via the technique of selective laser melting (SLM) for enhanced shear resistance relative to other fabrication processes such as vacuum brazing. Four-point bending responses of as-fabricated 3CSP specimens, including bending resistance and initial collapse modes, were experimentally measured. The bending characteristics of the 3CSP structure were further explored using a combined approach of analytical modeling and numerical simulation based on the method of finite elements (FE). Both the analytical and numerical predictions were validated against experimental measurements. Collapse mechanism maps of the 3CSP structure were subsequently constructed using the analytical model, with four collapse modes considered (face-sheet yielding, face-sheet buckling, core yielding, and core buckling), which were used to evaluate how its structural geometry affects its collapse initiation mode.

Influence of Backing Plate Support Conditions on Armor Ceramic Protection Efficiency.

The bilayer composite ceramic armor is widely used in the world, while the protection efficiency of the armor ceramic in it still confuses researchers. This study applied a numerical simulation method to produce a general equation that describes the relationship between the protection efficiency of the armor ceramic and the supporting conditions of the backing plate, thereby enhancing the current understanding of the composite ceramic armor. The results indicated that the protection efficiency of the armor ceramic can be divided into three parts: (1) the basic protection efficiency, (2) the increment efficiency caused by inertial support, and (3) the increment efficiency caused by mechanical support. The inertial support is related to the density and thickness of the backing plate, and the mechanical support is related to the elastic modulus and yield strength of the backing plate materials. The inertial support exhibited a positive correlation with the protection efficiency of the armor ceramic before it reached the Scr; after that, the protection efficiency of the armor ceramic would remain stable. In addition, the mechanical support showed a linear, positive correlation with the backing plate stress at ε0.