Real-time obstacle avoidance for manipulators and mobile robots
International Journal of Robotics Research
A neuron model with fluid properties for solving labyrinthian puzzle
Biological Cybernetics
An algorithm for planning collision-free paths among polyhedral obstacles
Communications of the ACM
Collision Detection between Robot Arms and People
Journal of Intelligent and Robotic Systems
Transport robot with network control system
IJCAI'75 Proceedings of the 4th international joint conference on Artificial intelligence - Volume 1
Neural Networks and Micromechanics
Neural Networks and Micromechanics
IEEE Transactions on Systems, Man, and Cybernetics, Part A: Systems and Humans
Permutation Coding Technique for Image Recognition Systems
IEEE Transactions on Neural Networks
| Hi-index | 0.00 |
When a robot moves among obstacles it sometimes needs to perform relatively complex maneuvers. The problem of selection of adequate maneuvers can be considered as an image recognition problem. At the input we have the image of a situation presented by the camera or rangefinder, and the output will present us with an appropriate maneuver that has to be performed to approach the goal. In contrast to the usual recognition system, the number of possible maneuvers can be enormous. It is practically impossible to enumerate them and give a name to each of them. For this reason it is necessary to develop a formalism for different maneuver representations. We propose the use of Hebbian ensemble neural networks for this purpose. This paper contains a brief description of Hebbian ensemble neural networks and some results of information capacity estimation. Information capacity shows how many ensembles can be stored in the neural network of a given size (given number of neurons in the network). It is shown that the number of ensembles can be much larger than the number of neurons in the network.