Over the past three decades, there has been considerable focus on determining path planning for mobile robots, aiming to find safe and efficient routes between starting points and destinations. Exploring various navigational techniques holds promise for broadening the scope of applications for autonomous mobile robots. This research introduces a novel navigation strategy by combining the Particle Swarm Optimization (PSO) method with the Bacteria Foraging Optimization Algorithm (BFO). Inspired by the foraging behavior of bacteria, particularly E. coli, this bio-inspired approach is utilized to optimize routes for mobile robots, harnessing the advantages of PSO. The primary objective is to establish routes that are both feasible and secure. The proposed model involves a robot that mimics bacteria behavior to identify the most optimal route within environments cluttered with obstacles, connecting an initial point to a designated destination. This approach is implemented and evaluated across various scenarios, demonstrating the effectiveness and potential of the proposed method.

Autonomous mobile robots are being increasingly used in various application areas such as smart manufacturing, warehouse automation, healthcare assistance, and intelligent urban service systems. In these applications, the robots are required to navigate in complex and dynamic environments while avoiding static and dynamic obstacles. In traditional navigation and path planning approaches, there are limitations in highly dynamic environments because of their rule-based or static assumption about the environment, thereby limiting their adaptability to real-time changes. This paper proposes a reinforcement learning-based navigation system for autonomous mobile robots in dynamic environments. The navigation problem is modelled as a Markov Decision Process (MDP), which allows the agent to learn optimal control actions by interacting with the environment. Based on sensor information such as obstacle distances and robot velocity, the agent produces navigation decisions to reach targets efficiently while avoiding collisions. The proposed system was tested in a simulation environment with multiple dynamic obstacles. The system's performance was evaluated using multiple parameters. The results indicate that the proposed system outperforms classical, optimization-based, and learning-based navigation techniques in terms of navigation efficiency and collision rate. The major contributions of the current research include: development of a reinforcement learning-based framework for navigating mobile robots through complex and dynamic environments, evaluating the framework in complex and dynamic environments, and comparing the framework with a wide range of navigation techniques. The results validate the efficiency of reinforcement learning in terms of increasing the intelligence of mobile robots in complex and dynamic environments.

Intrusion detection systems (IDS) are essential for protecting computer systems and networks against a wide range of cyber threats that continue to evolve over time. IDS are commonly categorized into two main types, each with its own strengths and limitations, such as difficulty in detecting previously unseen attacks and the tendency to generate high false positive rates. This paper presents a comprehensive survey and a conceptual overview of Hybrid IDS, which integrate signature-based and anomaly-based detection techniques to enhance attack detection capabilities. The survey examines recent research on Hybrid IDS, classifies existing models into functional categories, and discusses their advantages, limitations, and application domains, including financial systems, air traffic control, and social networks. In addition, recent trends in Hybrid IDS research, such as machine learning-based approaches and cloud-based deployments, are reviewed. Finally, this work outlines potential future research directions aimed at developing more cost-effective Hybrid IDS solutions with improved ability to detect emerging and sophisticated cyberattacks.

Real-time embedded vision systems are subject to severe constraints in terms of inference latency, memory usage, and power consumption. This makes the deployment of deep learning models a severe challenge. While deep learning models have been able to deliver high accuracy in a wide variety of computer vision-related tasks, the straightforward deployment of these models on embedded systems results in a high computational overhead. This paper aims to explore the trade-offs associated with the deployment of deep learning-based computer vision models on resource-constrained embedded systems. The performance of various optimization techniques, such as structured pruning, low-precision quantization, and hardware-aware inference, has been evaluated. A deep learning-based computer vision model for object detection has been implemented to simulate a realistic deployment environment through the TensorFlow Lite framework. The results of the experiment demonstrate that a significant improvement in the latency as well as the memory usage of the model can be achieved at the expense of a marginal accuracy penalty.