The accumulation of discarded plastics contributes significantly to white pollution and biomagnification, positioning biodegradable plastics as a promising alternative. Currently, bioplastics account for less than 1
Large language model (LLM) agents that interact with graphical user interfaces increasingly rely on either raw screenshots or platform-specific accessibility application programming interfaces (APIs) to perceive interface state. Both approaches have limitations for assistive applications: screenshot-based perception lacks the semantic roles and relationships required by screen readers, while platform-specific APIs such as Windows UI Automation, macOS Accessibility, Android AccessibilityService, and web ARIA require separate integrations for each platform. This paper proposes an architecture that uses the Model Context Protocol (MCP) as a unified transport and schema layer between heterogeneous accessibility frameworks and LLM-based assistive agents. An MCP accessibility server exposes ARIA-aligned roles, labels, states, and focusable-element hierarchies through a platform-independent representation, enabling consistent interaction across operating systems and applications. The framework also introduces an MCP resource model for persisting user accessibility preferences across sessions. The architecture is analyzed with respect to three research questions: protocol extensibility for accessibility-tree representation, latency and semantic fidelity trade-offs between accessibility trees and screenshot-based perception, and support for persistent accessibility profiles through MCP resources. Rather than presenting an empirical implementation, this work contributes a conceptual framework supported by comparative analysis of accessibility APIs, GUI agent architectures, and the MCP specification. The analysis suggests that a standardized MCP accessibility layer can reduce platform-specific integration complexity while preserving the semantic information required for accessible AI agents, providing a foundation for future implementation and evaluation.
HITL automation is increasingly used in finance to enhance the accuracy of decisions, operational efficiency, and regulation compliance. The use case scenarios include loan approval and fraud detection, implemented through Camunda BPM and Java-based UIs. The system has automated rule execution, DMN-based decision logic, and specific human verification; thus machine intelligence collaborates with human intelligence. Once we took it for a test drive, we realised it could save a lot of process time and human effort. It managed to automate 60-65% of the routine tasks while the accuracy for complex cases really improved with human moderation. In HITL workflows decisions are 15-20% better, with less errors compared to a fully automated process. Scalability testing shows a steady performance with high concurrency loads. The audit trail visualization offers a full traceability dimension in support of compliance. The results show that HITL automation is effective in doing important financial tasks by creating automation with human intervention.
An innovative approach for assessing partial speech problems is the automatic speech recognition system. It is a technology is used for significant quantity of naturally occurring speech from speakers with disabilities which can be subjected to linguistic and acoustical study. Kannada, an important Indian dialect, is the focus of the present study. ASR offers a variety of strategies based on the situation. The isolated word ASR system for the Kannada language is presented in the proposed study using artificial neural networks. Words that are spoken can be converted into the matching written formats using isolated word recognition. The properties of a speech stream are distinguished using Mel Frequency Cepstrum and Linear Predictive Coding (LPC). The objective of this study is to create a standalone word recognizer utilizing a word acoustic model and a combination of LPC and ANN. The proposed system consists of the two stages they are training and testing phases. 50 isolated words are recorded utilizing multiple speakers in a silent environment as part of the training process. There were both individuals with partial speech disorders and normal speakers of all ages and genders among individuals. The features of the resulting sample are extracted using LPC and trained using ANN after each word is repeated several times by the particular speaker. During the testing phase, the system receives input of isolated word utterances, LPC coefficients are determined, and recognition is done using the sigmoid function and text that corresponds to the stated word. The system was trained and tested using 50 isolated words from the Kannada language. The system results had a 95.8
In this work, a Single-Ended Primary Inductor Converter (SEPIC) with improved power factor correction stage is developed for Electric Vehicle (EV) charging application. Practically, it is desirable to employ a Power Factor Correction (PFC) circuit that operates in a way that input current is taken from the source remains continuous, power factor is high, step-down operation, and stable output regulation. This research focuses on the analysis and design of a non-isolated hard switching Discontinuous Conduction Mode (DCM) SEPIC PFC, which addresses these requirements. By operating in the DCM mode, the SEPIC converter enables variable step-down output voltage, continuous conduction of input current, and eliminates reverse recovery losses simultaneously. The paper begins with circuit analysis for both DC–DC conversion and PFC operation and transfer function for PI controller can be derived based on averaged switching model, small signal model. The close loop controller is implemented in voltage follower mode, for the proposed DCM operated SEPIC PFC. The PFC stage is implemented in MATLAB Simulink environment and results obtained to validate the concept of inherent Power Factor Correction using DCM converter operated in voltage follower mode of control.
