Development of software for mobile robot control over a radio frequency communications link

In the spring of 2001, work began on a mobile robot for an advanced computer engineering class. One of the key parts to the success of this robot would be the ability to control it from a computer terminal. In order to make this feasible, the computer would communicate to the robot through a wireless connection. The robot's on-board computer would perform any computations needed, so the robot would not need to send any information on the controlling computer. The communication link was designated to be one way, from the computer to the robot. For reasons including cost efficiency, ease of use, and range, RF communication was chosen. Using two development boards from Linx Technologies, we had the capability of communicating through the computer's serial port. The boards boasted a line of sight range of about 1000 feet. The board acts as nothing more then a transmitter and receiver, so any error correction must be dealt with in the software. Often errors come from interference created by surrounding RF devices, and appear as incorrect bits in the data stream. In order to handle such errors, methods involving redundancy were used. When the stream of information is sent, it is initialized by a start byte. This is followed by the commands, each of which is repeated three times followed by a stop byte. On the receiving end, the software waits for the start byte and then stores each byte that follows. They are then compared with one another and determined to be either clean data or an error. If the data is clean, the robot then executes the commands. By monitoring the data received by the receiver we were able to see the need for such software. Upon powering up the board, miscellaneous bits are received, and if there is no error correction, the robot could misinterpret these for a command. The boards also received stray bytes while idle, so the software would also eliminate this. Software development was started and initially tested over a crossover cable connecting the two serial ports. This allowed us to ensure that the software was working with out seeing any errors at first. Finally the software was used with the RF boards and seen to be functional. This paper explores, in depth, the methods used by software to correct errors that may develop in RF communication. The materials used are discussed in greater detail along with the creation of the software. Experimental results show the methods to be precise, and fit the design application needs.