The Nomad 200 is an integrated mobile robot system with four sensory modules including a tactile module consisting of 20 pressure sensitive sensors, an infrared module consisting of 16 sensors providing a 360 degrees environment coverage and a range up to 24 inches, an ultrasonic module consisting of a 16 channel ranging system providing range information from 17 inches (42 centimeters) up to 255 inches (6.5 meters), and a structured light vision system which through a horizontal light projection estimates the range to the object using triangulation. This roving robot is enhanced here with the integration of CCD cameras for augmented computer vision applications. The Nomad robot has on-board processors dedicated to the control of the sensors and the robot's motor. The Nomad 200 robot is about 35 inches high (close to 1 meter) with a radius of about 20 inches. Its mobility is achieved through a three-wheel base which can translate and rotate with a translational speed of 20 inches per second and a rotational speed of 60 degrees per second.
 

Projects using the Nomad 200:

1.  Speech Recognition

The goal of this part of the project is to take a recognized spoken word or written command and send the corresponding instructions to be executed by the Nomad 200. This is done via a C++ program that takes a pseudo-instruction and transforms it into the required syntax and parameters needed by the robot for execution. Available instructions include: go straight indefinitely until an obstacle is encountered, turn left or right for a determined number of degrees, turn for 180 degrees, go forward for a specified distance, and go backwards for a specified distance. Currently the program accepts mnemonics identifying a desired command as well as the required parameters, such as speed and distance, and directs the robot to execute the command remotely from a workstation. This program makes use of the front and rear infrared and sonar sensors of the Nomad 200 to recognize when obstacles are near. When this is the case, the program instructs the robot to cancel the execution of the current command. In the future the command execution program will be combined with the speech recognition one to accept either spoken (via a microphone) or written (via the keyboard) commands. A default section was added to this program to ensure that a command would be properly executed even when incomplete written mnemonics are issued. In this case, intermediate default translational and rotational velocities are assumed.
 

2.  Two-dimensional plot of the Real Environment using the sonar and infrared sensors

    The first stage of integrating the hardware and the software of the Nomad 200 robot was the development of an algorithm to display a two dimensional plot using the sonar sensors. The algorithm was tested using the simulated echo response of the sonar sensors.  This algorithm was implemented in a Nomadic Host Software Development Environment (Version 2.6.2) using C++ programming language. The x and y coordinates of the simulated objects detected from the echo of the sonar sensors were calculated and a two-dimensional plot was then displayed on the simulated environment of the Nserver’s Graphic User Interface (GUI). If the sonar’s sensor echo was at maximum distance, then no object was assumed detected.  Therefore, that coordinate was not displayed on the Nserver’s GUI.

    The second stage will be in testing the developed algorithm in a real environment using the real robot.  The real environment involves the detection of obstacles, where noisy measurements exist.  Therefore, the robot must decipher the difference between ‘noise’ and actual objects.  For this stage, the sonar sensors with the addition of the infrared sensors will be taken into account.  The sonar sensors transmit a signal for long distances that calculates the time it takes the echo of the surrounding objects to return.  The infrared sensors work in the same way, but the signals detect a shorter distance, therefore the signal will have less noise corruption.  These echoes of the surrounding objects are then converted to tens of inches.  The use of the infrared sensor is expected to improve the accuracy of the detection of noise.