Object detection, 3-D coordinate extraction and pose estimation for an autonomous medical robot

Abstract

The presence of autonomous robots is increasing in every field by the day. Robots require computer vision to interact with the surrounding environment. Computer vision is an area of study which deals with the algorithms which interpret the data from images or videos. One scenario in which robots could be helpful is in the medical field. This paper presents a set of algorithms which cater to the needs of a medical robot. An object detection algorithm is trained for the objects expected in medical environment. The detected object's 3-D coordinates are extracted. To find the 2-D orientation of the object a novel approach based on edge detection and linear regression is described and implemented. These algorithms help the robots to serve in the medical environment where the contagious diseases are a threat to healthcare workers as seen in the COVID-19 pandemic.