Perception & Behaviour:   To resemble human emotion we have introduced a new kind of Unsupervised learning process, which we have developed recently. This was developed based on some observation on child psychology. As a result, the robot can be trained with different soccer tactics, and we are predicting that it will evolve on a small scale. This process resembles human brain in a way that, it consists of numerous simple processing units which will do similar processings and will work in parallel to perform a specific job. Its resembles artificial neural networks in a way that this network can also be trained, but it it also differs in another way. As there is a chance of overfitting inputs in case of artificial neural networks, hence diffferent algorithms like backpropagation algorithm is used to minimize error in each learning stage in such a way that overfitting doesnot occur. But our network used here doesn't depend on fitting an equation, as a result the question of overfitting is impossible.
This link contains a file which will illustrate how inverse kinematics will be used for different joints
This link contains a file which will illustrate our observations on human psychology and how the whole process will work in parallel
Hosted by