S. Zrehen and P. Gaussier. Why topological maps are useful for learning in an autonomous agent. In J.D. Nicoud and P. Gaussier, editors, From perception to action conference, Los Alamitos, CA, 1994. IEEE Press. 24  Home/Search   Document Details and Download   Summary   Related Articles   Check

....the robot and the obstacles. What can one do with a robot that has 8 not very accurate sensors Apply fuzzy logic It can be one of the ideas. There are some others: genetic algorithms [Flo94] subsumption architecture [Bro86] distributed adaptive control [Ver92] learning with neural networks [Gau93, Zre94], artificial intelligence approaches [Bas93] In next section, we will describe a very simple fuzzy controller for this robot. 3.1 A simple fuzzy controller for obstacle avoidance Suppose that the first goal is to do obstacle avoidance with a mobile robot. Whatever the mobile robot has to ....

S. Zrehen and Ph. Gaussier. Why topological maps are useful for learning in an autonomous agent. In Proceedings PerAc, Lausanne, September 1994. IEEE Press.

....and neuro fuzzy control 13 What can one do with a robot that has 8 not very accurate sensors Apply fuzzy logic It can be one of the ideas. There are some others: genetic algorithms [Flo94] subsumption architecture [Bro86] distributed adaptive control [Ver92] learning with neural networks [Gau93, Zre94], artificial intelligence approaches [Bas93] In next section, we will describe a very simple fuzzy controller for this robot. 3.1 A simple fuzzy controller for obstacle avoidance Suppose that the first goal is to do obstacle avoidance with a mobile robot. Whatever the mobile robot has to ....

S. Zrehen and Ph. Gaussier. Why topological maps are useful for learning in an autonomous agent. In Proceedings PerAc, Lausanne, September 1994. IEEE Press.

....the robot and the obstacles. What can one do with a robot that has 8 not very accurate sensors Apply fuzzy logic It can be one of the ideas. There are some others: genetic algorithms [Flo94] subsumption architecture [Bro86] distributed adaptive control [Ver92] learning with neural networks [Gau94, Zre94], artificial intelligence approaches [Bas93] The simple fuzzy controller for the obstacle avoidance for Khepera is described in details in [God95] The controller has 4 inputs and 2 outputs as shown in Figure 5. The robot avoids obstacles very successfully. The critical problem is that the number ....

S. Zrehen and Ph. Gaussier. Why topological maps are useful for learning in an autonomous agent. In Proceedings PerAc, Lausanne, September 1994. IEEE Press.

....information and cannot take advantage of what has been learned by other neurons. On the contrary, with topological preservation, a neuron in the neighborhood of a given winner will respond accordingly depending on its physical distance to it. It thus gives interesting generalization capabilities [39]. The PTM algorithm main characteristics are listed below: ffl Input patterns and synaptic weights take only binary values ffl A new pattern has to be learned immediately (incremental learning) ffl A selectivity function is used to allow neurons which have stored a shape to have a higher ....

S. Zrehen and P. Gaussier. Why topological maps are useful for learning in an autonomous agent. In J.D. Nicoud and P. Gaussier, editors, From perception to action conference, Los Alamitos, CA, 1994. IEEE Press. 24

....give any information and cannot take advantage of what has been learned by other neurons. On the contrary, with topological preservation, a neuron in the neighborhood of a given winner will further respond depending on the distance to it. It thus gives interesting generalization capabilities [68]. The PTM algorithm main characteristics are listed below: ffl Input patterns and synaptic weights take only binary values ffl A new pattern has to be learned immediately (incremental learning) ffl A selectivity function is used to allow neurons which have stored a shape to have an higher ....

S. Zrehen and P. Gaussier. Why topological maps are useful for learning in an autonomous agent. In From perception to action, 1994.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC