Abstract of project:
The main objective of the project is to simulate the behavior of a pet robot. The interaction of pet with the user is implemented through speech. The aim is to make the pet learn sequence of commands given by user. Here four basic commands have been used to train the pet initially viz. ‘Move’, ‘Attack’, ‘Twist’, ‘Neon’. ‘Move’ specifies the forward movement; ‘Attack’ characterizes the rotation of its arm. It turns right to react on ‘Twist’ command while it beeps identifying its name ‘Neon’. Neon learns to perform a sequence of any of the three commands at a time. To implement the learning we have used the Artificial Immune System (AIS). The basic fundamental of AIS is to immune the system by capturing and thus suppressing the Antigens using the set of Antibodies. The learning process is also influenced by Reward/ Penalty given by user. It was observed that the learning procedure using AIS takes much less time than any other learning mechanism. The energy level increases with intensity of Reward given and decreases with the actions performed. The tradeoffs are maintained by checking the critical value, below which the motor speed is affected. The learning is implemented in two levels of hierarchy. When the first level of learning of sequence of commands is over, Neon fine tunes itself, in the second level, to adapt to attributes of different operations depending on different user profiles.