Adaptive Pattern Recognition of Object for Autonomous Robot

Abstract

Autonomous robots can execute desired tasks in an unstructured environment without newlinecontinuous human guidance. An adaptive pattern recognition and pick and place of object newlineusing autonomous mobile robot is a complex task. The designed robot is required to newlinerecognize object, identify grasp location and size, pick and place, escape from collision newlinethreat and monitor light intensity, temperature, and battery voltage. Several types of robots newlinehave some degree of autonomy in different ways. Domestic robot is independent in their newlinerestricted work environment. Home robotsand#8223; workspace is difficult and can often enter newlineconfused unexpected variables. As robot is autonomous, it is proposed to find out its energy newlinerequirement that is stored in battery. It has to charge to the required level by itself, newlinewhenever charge left inside the battery is low. It is proposed that, the robot has to find out if newlinethe temperature of the motor driver is above the pre-determined level, fan should be turned newlineon to cool the motor drivers. If the intensity of the surrounding light of robot environment newlinereaches below a threshold level, then LEDs should be turned on. newlineWe demonstrate machine vision application for autonomous mobile robot to pick and place newlineobject in an indoor home environment. A miniature robot is developed and automated using newlinemicrocontroller ATmega32 and NI LabVIEW. Information passes between ATmega32 newlinemicrocontroller and notebook pc via RS232 cable and VISA. Machine vision part is done by newlineNI Vision of National Instrument. newlineRecognition of object and grasp location is adept by pattern matching algorithm of IMAQ newlinevision. Pattern image templates are created and stored in memory as template database. newlineWhen the application runs, the camera captures live image from its environment into the newlineapplication. Vision part of the system examines the image and sends corresponding newlineelectrical signal to robot control for executing grasp and release of object, after identifying newlineit. IMAQ algorithm estimates gripper configuration of open/close size ba