Manipulator robot with artificial vision

Abstract

This paper presents the partial results achieved in the project for the construction of a Robot Manipulator and the design of applications to provide it with Artificial Vision capabilities by means of a video camera for fruit and seed operations in the agricultural and food industry. The scope of this article is its construction. Among other options, we chose a DAUERHAFT brand structure that includes the servomotors, of which only those that represent 3 DOF are left and the others are disabled. The clamp is also left. The joints that characterize the operations of turning (roll), elevation (pitch) and deviation (yaw) are left, in similarity with the anatomy of the human being: the links of the trunk or waist, of the shoulder and of the elbow; the reference is the robot Cincinnati Milacron T3 and the PUMA robot from Unimation. The chosen digital system is a master-slave arrangement in which the master is a PC laptop computer programmable in Python language with USB interface to the slave, an Arduino Mega development board that with a support voltage source connects directly with the servomotors and the potentiometers. In each of the three joints, gear systems that transmit the movement to 100kOhm potentiometers were adapted so that, through a displacement-voltage relationship, the position of each link is obtained. Through an interface in Python, individual manipulation of each of the servomotors is possible.

The target position is provided and smoothly the mechanism reaches its end point. Image processing training for color, object and contour recognition is also done in isolation. The second part of this project, the future work in progress; is to unite these two subsystems and the development of applications through movements with inverse kinematics.