The necessity of ensuring people's safety in urban conditions during gathering in a certain indoor or outdoor area, as well as during the operation of technical equipment, requires measurement of their number, weights, the distance between them, density and many other parameters. The collection of those parameters and analysis allows you to ensure the optimal numerical value of people per area and not exceed the critical peak value, ensuring the safety of people's health and life. However, automating those processes requires stable network bandwidth and enormous computing resources for data transferring to the data center and its processing, as well as applying artificial intelligence algorithms to the data and sending the results back to the end-user. In this paper, to avoid the disadvantages of cloud computing, we consider Raspberry Pi with a camera module as an edge device to develop an automated people-counting system based on using OpenCV algorithms and Edge Computing for Safety Management. The proposed system can be used for indoor or outdoor areas, during scaffolding works and lift's operation for controlling an optimal number of people. As a result of the application the edge computing, network bandwidth and computing resources of the data center have been optimized.

Maintaining a specific geometric formation during the movement is crucial for multiagent systems of mobile robots in various applications. Proper coordination can lead to reduced system costs, increased reliability and efficiency, and system adaptability and flexibility. This research proposes a novel movement coordination method for self-governing multiagent systems of intelligent mobile robots. The proposed method uses a leader-follower technique with a virtual leader to maintain a specific geometric structure. Additionally, the epsilon greedy algorithm is utilized to avoid loops. To reduce power consumption, it is proposed to turn on only a few robots' lidars at a time. They could drive all the robots in the group, allowing them to reach the goal without colliding with obstacles. Experiments on a complex map with nine robots were conducted to test the method's effectiveness. The success rate of the swarm reaching the target position and the number of steps needed were evaluated. Testing varied angular velocities of 1 to 20 degrees and linear velocities of 0.1 to 5.5 m/s. Results show the method effectively guides the robots without collisions. This method enables a group of self-governing multiagent systems of intelligent mobile robots to maintain a desired formation while avoiding obstacles and reducing power consumption. The results of the experimental study demonstrate the method's potential to be implemented in real-world missions and traffic management systems to increase efficiency and reduce costs. The proposed method can be utilized in military missions and traffic management systems, where maintaining a specific geometric formation is crucial. The method's ability to avoid obstacles and reduce power consumption can also lead to reduced costs and increased efficiency.

The task of keeping a specific geometric configuration while following a designated path is common in various fields that involve autonomous robots. When done correctly, this approach can result in several advantages, including a reduction in system costs and an increase in system reliability and efficiency, while maintaining the adaptability and flexibility of the system. The concept of maintaining a particular geometric pattern during motion is frequently utilized in different scenarios. For example, unmanned military vehicles must maintain a particular formation to explore and cover terrain during missions. Another example is in “Smart Highways,” where the traffic system's capacity can be enhanced significantly by having vehicles move in groups at the same speed while maintaining a certain distance between them. This research presents an algorithm that allows a group of self-governing mobile robots to move while maintaining a specific geometric structure. A behavior-based control algorithm is advanced to regulate the movement of these systems. The coordination of the robots within the system is accomplished through the implementation of a leader-follower technique, with virtual leader.

In this paper, a new encryption algorithm Twine-Mersenne was developed based on the Twine algorithm with the addition of a random number generator for the dynamic generation of S-boxes. Dynamic generation of random numbers based on the Mersenne Twister helps to increase the cryptographic strength of the proposed algorithm. The algorithm we propose solves the issues of optimizing the costs of computing and energy resources of internet of things (IoT) devices, using a combination of lightweight cryptographic principles and fuzzy logic, and also provides reliable security and intelligent authentication of the mobile application user. The paper also considers the practical implementation of the proposed algorithm based on Arduino ESP32, a device with limited computing resources. In addition to this, fuzzy logic has found its practical application in the field of intelligent user authentication in developed mobile applications based on Arduino Studio for mobile cellular applications. As a result, the proposed lightweight encryption algorithm has proven itself to be an effective tool in ensuring the security of data transmission and communication between IoT devices.

Background: Coordinating differential-drive mobile robots for landmark coverage is challenging due to non-holonomic dynamics, clutter, and sparse rewards. Standard multi-agent RL pipelines often show unstable learning and inconsistent completion in this setting. Methodology: We adopt a centralized-training, decentralized-execution actor-critic without inter-agent commu nication. Our replay-centric design combines a tagged buffer that up-samples goal-reaching transitions and an offline replay initialization that seeds early learning with curated trajectories. Dynamic task assignment uses the Hungarian algorithm during training and evaluation, and we benchmark against uniform replay and established variants. Results: In a cluttered six-robot arena, the approach improves training stability relative to uniform replay. Convergence is faster and requires fewer updates to reach consistent success. Coverage efficiency increases as landmarks are reached earlier across runs. Collisions per episode decrease without adding communication or architectural changes. Multi seed evaluations show gains that persist with narrow confidence intervals. Train-evaluation gaps shrink on unseen maps, indicating improved generalization. Ablations attribute complementary ben efits to the tagged and offline components. Performance remains competitive with prioritized and hindsight replay baselines under matched budgets. Computational overhead is small because sam pling logic changes while network sizes do not. Conclusions: Focus ing on replay design substantially stabilizes multi-agent learning for differential-drive coordination. The pipeline integrates cleanly with standard CTDE implementations and supports practical deployment in coverage tasks.