Investigating Collision Detection Techniques in Six-Degree-of-Freedom Collaborative Robots

Abstract

In the contemporary era of advanced technology, Collaborative Robots, known as Cobots, have emerged as a highly promising domain of research and application. Cobots represent a category of robots endowed with the capacity for direct interaction with human counterparts within shared working environments. Their design philosophy centers around harnessing the synergies between human and robotic capabilities, thereby augmenting work efficiency while concurrently ensuring a secure and productive work environment. A pivotal facet of Cobots pertains to their innate ability to operate in a secure and human-friendly manner. This is achieved through the implementation of autonomous collision detection mechanisms, enabling immediate cessation of operation to mitigate potential harm to humans. This attribute assumes particular significance when Cobots and humans collaborate within the same physical workspace. Our research endeavors are concentrated on the enhancement of performance and reliability within Cobots' collision detection systems. To this end, we propose the utilization of two supervised machine learning methodologies, specifically Support Vector Machine Regression (SVMR) and 1D Convolutional Neural Network (1D CNN), to bolster the precision and speed of collision detection for the CURA6 robotic arm- based on Intema's CURA6 dataset. The findings of this study are poised to significantly augment the operational capabilities of Cobots, thereby reducing the risk of accidents in industrial and manufacturing settings.