Enhancing Robotic Assistance through Advanced Human Activity Recognition: Pioneering Human-Robot Interaction and Intelligent Automation

Abstract

Robotic assistance is, therefore, the use of a robot in tasks to complement human input in activities from simple household work to critical surgeries, making work to be more effective, accurate, and secure in medical, automotive, and household establishments. With these tasks being executed with robots, it is crucial that these robot understand and distinguish human movements in order capture their behaviors correctly. This research meets this need through proposing a novel advanced Automatic Human Activity Recognition system specialized for improving robotic support. The procedure of the study starts with the preparation of video data that include the formation of video frame batches and the depictions of the class distributions in training and testing sets. Such preparation helps to feed the deep learning model with good quality well-arranged data into the model for training. The model features a dual-component architecture: A feature extractor that is able to analyze video sequences and a classifier that is able to sort activities. Following the training, there is always evaluation of the model using a confusion matrix to check the level of accuracy in the model. Also, the real-time video analysis is possible and the predictions can be made immediately while they will be saved to the CSV file and then find the most frequently occurred activities. This research augments the human activity recognition in a way that robots become intelligent and perceptive, making their performance improve in getting assistance to humans in numerous ways.