In the course of this project we fabricated devices that were differentially steered as well as conventional car-like devices; we also developed mathematical models and performed experiments on both of them. The differentially steered mechanisms are the simplest kind where the two wheels on either side are given different torques in order to steer the vehicle; most commercial “robots” are of this type. They are navigable in very narrow environments and are easy to implement. However, it can be shown that certain motions are kinematically not possible, they are extremely sensitive to relative velocities, and hence, they are sensitive to slightest variations in the ground. Hence, they are highly inaccurate in terms of paths that we would like them to follow.

The conventional car-like devices are driven and steered with different actuators. They are inherently more accurate. However, in order to drive them accurately, they should not slip. This leads to nonintegrable differential equations that make the analysis and control quite challenging. In addition, as is obvious from the practical problem of parallel parking of a car, instantaneous motions are reduced in car-like devices.

A third kind of device that was fabricated was a combination robot that is driven with two independently controlled DC motors installed on the front axle. It has a single steering wheel at the rear axle being positioned by a servomotor. The velocity difference of the driving wheels is actively controlled based on the angle of the steering wheel. This reduces the slip of driving wheels and results in a very accurate position and path-tracking control compared to conventional car-like robots, in which the velocity difference of the driving wheels is passively controlled based on the friction between wheels and the ground. The car-like robot has an infrared sensor in the front, which is installed on a servomotor. With the help of the servomotor, this infrared sensor is able to rotate 180 degrees and scan the environment. This robot has been used for implementing obstacle avoidance theories.

The Mule

A mobile platform called “The Mule” was especially designed for outdoor situations. Its innovative designed drive train consists of tracks and pulleys, which give the robots good traction and performance on rough surfaces. The track can be tilted with respect to the Mule chasse, which controls the ground clearance and stability of the Mule on the inclined surfaces. The platform contains a controller, power source, sensory elements and the electrical portion of the actuation (motors). The volume of the Mule is roughly 560 cubic cm (34 cubic inches). Two independently controlled motors (from MicroMo^TM) drive the Mule. They are Neodymium, permanent, magnet coreless DC motors with build-in 2mm wide magnetic encoder and gear head. The power source is a Nickel-Metal Hydride 9.6 V, 1100 mAh battery, which can drive the Mule for one hour in full power. The maximum speed of the Mule is approximately 0.3 m/s on a horizontal smooth surface. The controller used was the Eyebot MK3, developed by researchers at Western Australian University.

Real device implementation

A real robot has to do many things at the same time (like a human being). Hence, multi-tasking is an essential feature of such a device. We developed an algorithm to carry out the different tasks by time-sharing the microcontroller. The main behaviour modes are as follows.

• Cruise Behavior – the robot has to keep cruising in accordance with the preset paths (or dynamically determined ones); this would involve steering in accordance with the desired paths.
• Avoid Behavior – the robot must sense the presence of static or moving obstacles with its sensors (IR), and avoid them.
• Escape Behavior – if the robot collides with anything unexpectedly (detected by the bumper sensors), it should recover before damage occurs.
• Drive Motor Behavior – the controller software must provide the right currents to drive the drive motors with the necessary torque.

An arbitration routine rules over the above behavior modes and ensures that all the criteria are met simultaneously.