Ensuring safety while retaining maximum performance is a basic requirement for automatic cyber-physical systems, especially for safety-critical applications. A quadratic programming optimization framework called MPC-CBF has recently been presented, which directly unifies model predictive control (MPC) with control barrier functions (CBFs) over the prediction time horizon. However, the conservative nature of CBFs can lead to feasibility problems in real applications. Based on the analysis of the role of the decay rate and the conservative accumulation phenomenon in standard CBF formulations, this paper proposes to directly optimize CBF constraints within the MPC framework. By regarding CBFs as a safety restriction level indicator and an optimizable constraint within the MPC framework, the trade-off between feasibility and safety can be adaptively optimized. The proposed Optimizable CBF (OCBF) model removes the hyper-parameters selection problem in standard CBFs and can adaptively adjust the safety restriction level and increase behavior diversity by adding the corresponding objects in the cost function in MPC. To eliminate the accumulation effects of actual values of the CBF constraints in previous time steps, this paper further proposes a General OCBF (GOCBF) formulation. Compared with existing formulations, the safety margin defined in our GOCBF has intuitive physical meanings and thus provides a more flexible and intuitive mechanism to compromise different objects in terms of ensuring safety while not undermining basic feasibility. Experimental results demonstrate that our algorithm provides a more flexible and intuitive mechanism to achieve this, thus improving feasibility and adding behavior diversity in the MPC-CBF framework.

Efficient and stable exploration remains a key challenge for deep reinforcement learning (DRL) operating in high-dimensional action and state spaces. Recently, a more promising approach by combining the exploration in the action space with the exploration in the parameters space has been proposed to get the best of both methods. In this article, we propose a new iterative and close-loop framework by combining the evolutionary algorithm (EA), which does explorations in a gradient-free manner directly in the parameters space with an actor-critic, and the deep deterministic policy gradient (DDPG) reinforcement learning algorithm, which does explorations in a gradient-based manner in the action space to make these two methods cooperate in a more balanced and efficient way. In our framework, the policies represented by the EA population (the parametric perturbation part) can evolve in a guided manner by utilizing the gradient information provided by the DDPG and the policy gradient part (DDPG) is used only as a fine-tuning tool for the best individual in the EA population to improve the sample efficiency. In particular, we propose a criterion to determine the training steps required for the DDPG to ensure that useful gradient information can be generated from the EA generated samples and the DDPG and EA part can work together in a more balanced way during each generation. Furthermore, within the DDPG part, our algorithm can flexibly switch between fine-tuning the same previous RL-Actor and fine-tuning a new one generated by the EA according to different situations to further improve the efficiency. Experiments on a range of challenging continuous control benchmarks demonstrate that our algorithm outperforms related works and offers a satisfactory trade-off between stability and sample efficiency.

Unmanned aerial vehicle mobile ad hoc networks (UAVMANETs) formed by multi-UAV self-assembling networks have rapidly developed and been widely used in many industries in recent years. However, UAVMANETs suffer from the problems of complicated key negotiations and the difficult authentication of members’ identities during key negotiations. To address these problems, this paper simplifies the authentication process by introducing a Latin square to improve the process of signature aggregation in the Boneh–Lynn–Shacham (BLS) signature scheme and to aggregate the keys negotiated via the elliptic-curve Diffie–Hellman (ECDH) protocol into new keys. As shown through security analysis and simulations, this scheme improves the efficiency of UAVMANET authentication and key negotiation while satisfying security requirements.

Recently, control barrier functions (CBFs), are becoming popular and widely used as a safety-filter of the controller through quadratic programming(QP) optimization framework to ensure the forward invariance of the dynamic systems. However, for sequential safety tasks, the transitions of CBFs are required. To avoid control input discontinuities, firstly, a smooth and speed adjustable transition mechanism by adding a time-varying coefficient of different CBFs is proposed. Moreover, the value of the previous CBF constraint is fixed to reduce the change of the combined CBF constraint. Secondly, a method to integrate different sequential CBFs into a synthetic new one is presented. Through these methods, the smoothness of optimized control inputs can be ensured compared with existing formulations. Simulation results demonstrate that our algorithm provides a smoother transition mechanism to switch different CBFs, and improves the control performance consequently.