Department of Computer Science and Electrical Engineering
Department of Computer Science and Electrical Engineering
The first task of the thesis was to recreate RoboVac to working condition. This involoved designing and constructing a new main PCB and sonar sensors, and implementing basic software functions to control the stepper motors, sonar sensor s, bump sensors and vacuum cleaner. The second task, which was the main objective of this thesis, was to develop a control system that would enable RoboVac to efficiently cover the floor surface of a room.
The control system developed for RoboVac uses a behavioural control approach, rather than the planning approach used in most published work in the surface coverage field. Behavioural control approaches differ from planning approa ches in that they do not use a centralised controller or an internal world map representation. Instead, they use a number of cooperating behaviours, each of which handles a particular aspect of the overall task. In RoboVac, these behaviours interac t and control the robot using an activation blackboard – a region of data which holds the sensor data, motor commands and behaviour activation flags.
With two out of five designed behaviours implemented, the robot has been shown to cover approximately two-thirds of the floor area of an empty room, the missed area mainly due to robot dynamics and the sonar sensor limitations. Implemen ting the remaining designed behaviours would enable the robot to cover the floor area of a room containing obstacles, without missing areas to one side of the obstacle. With the addition of extra suggested behaviours, RoboVac’s coverage performance could be increased to nearly 100% of the floor area of a room containing obstacles. With considerations given to the design of the RoboVac body, RoboVac will soon be vacuuming the floors of real houses.
Complete thesis:
thesis.pdf
Conference paper: