Motion planning is one of the most critical problems in multirobot systems. The basic target is to generate a collision-free trajectory for each robot from its initial position to the target position. In this paper, we study the trajectory planning for the multirobot systems operating in unstructured and changing environments. Each robot is equipped with some sensors of limited sensing ranges. We propose a fully distributed approach to planning trajectories for such systems. It combines the model predictive control (MPC) strategy and the incremental sequential convex programming (iSCP) method. The MPC framework is applied to detect the local running environment real-timely with the concept of receding horizon. For each robot, a nonlinear programming is built in its current prediction horizon. To construct its own optimization problem, a robot first needs to communicate with its neighbors to retrieve their current states. Then, the robot predicts the neighbors' future positions in the current horizon and constructs the problem without waiting for the prediction information from its neighbors. At last, each robot solves its problem independently via the iSCP method such that the robot can move autonomously. The proposed method is polynomial in its computational complexity.

A self-adaptive software system is one that can autonomously modify its behavior at runtime in response to changes in the system and its environment. It is a challenge to model such a kind of systems since it is hard to predict runtime environmental changes at the design phase. In this paper, a formal model called intelligent Petri net (I-PN) is proposed to model a self-adaptive software system. I-PN is formed by incorporating fuzzy rules to a regular Petri net. The proposed net has the following advantages. 1) Since fuzzy rules can express the behavior of a system in an interpretable way and their variables can be reconfigured by the runtime data, the proposed model can model runtime environment and system behavior. 2) Since a fuzzy inference system with well-defined semantics can be used in a complementary way with other model languages for the analysis, thus the proposed model can be analyzed, even though it is described in two different languages: component behaviors in Petri nets while logic control in fuzzy rules. 3) The proposed model has self-adaption ability and can make adaptive decisions at runtime with the help of fuzzy inference reasoning. We adopt a manufacturing system to show the feasibility of the proposed model.

The Whale Optimization Algorithm (WOA) is a swarm intelligence algorithm based on natural heuristics, which has gained considerable attention from researchers and engineers. However, WOA still has some limitations, including limited global search efficiency and a slow convergence rate. To address these issues, this paper presents an improved whale optimization algorithm with multiple strategies, called Dynamic Gain-Sharing Whale Optimization Algorithm (DGSWOA). Specifically, a Sine–Tent–Cosine map is first adopted to more effectively initialize the population, ensuring a more uniform distribution of individuals across the search space. Then, a gaining–sharing knowledge based algorithm is used to enhance global search capability and avoid falling into a local optimum. Finally, to increase the diversity of solutions, Dynamic Opposition-Based Learning is incorporated for population updating. The effectiveness of our approach is evaluated through comparative experiments on blackbox optimization benchmarking and two engineering application problems. The experimental results suggest that the proposed method is competitive in terms of solution quality and convergence speed in most cases.

Regression testing plays an indispensable role in software maintenance, which refers to retest the software following modifications to determine whether the changes have introduced new faults. However, regression testing requires massive amounts of effort to achieve a high fault detection rate. To address this issue, the test case prioritization technique is used to improve the fault detection rate by adjusting the execution order of test cases. For model-based regression test case prioritization, existing approaches have been developed using the single aspect of model-related information extracted from the previous executed test cases. In this paper, a novel learn-to-rank technique is proposed to prioritize test cases by combining the multidimensional features of Extended Finite State Machine (EFSM) under test to improve fault detection rate. Specifically, our method utilizes the random forest algorithm to combine multiple existing heuristic prioritization methods. Detailed experiments are conducted to evaluate the proposed method's performance in terms of Average Percentage Fault Detected (APFD). The experimental results show that the mean APFD value of our method reaches 0.884 for five subject EFSMs, which is 33.9% higher than the compared methods.

Forecasting emergency calls is of great importance in practice. By forecasting the occurrence of unfortunate events, we can learn from these events and further prevent their occurrence in the future. However, because of the uncertainty of event occurrences, it is hard to guarantee their prediction accuracy. In this paper, a combined model, which consists of two parts, is proposed. The first part is a Poisson neural network model (PNN). It is responsible for basic forecasting, and its initial weights and thresholds are trained by applying a genetic algorithm. The second part consists of multiple linear regression (MLR), autoregressive integrated moving average (ARIMA) and multivariate gray (GM), which are responsible for estimating residual errors. The basic prediction result adjusted by the residual error is used as the final forecasting result. The proposed model fully takes the advantages of PNN, MLR, ARIMA and GM, and thus improves forecasting performance. Our method has been applied to the emergency calls of Ningbo, China. The experimental results show that the proposed model has advantages over some existing forecasting models, such as a neural network model, poisson regression and stochastic configuration networks in terms of Mean Absolute Percentage Error.