Learning systems and methods for market-based control of smart matter
AssigneeXerox Corporation
InventorsWarren B. Jackson, Oliver P. Guenther, Tad H. Hogg, Bernardo A. Huberman
Life of this patent
Estimating the term…
Abstract
A market based learning mechanism for controlling smart matter learn an appropriate organizational control structure for a multiple actuator-sensor dynamical system for use in optimizing system control. The more successful one or more of a plurality of control methods are in predicting system operation, the more weight the successful methods are given in controlling the system.
Claims18
A method for dynamically controlling of a multiple actuator-sensor smart matter dynamic control system, comprising:
- predicting future behavior of the multiple actuator-sensor smart matter dynamic control system using a plurality of control system models;
- determining at least one control system model which is more successful than at least one other model of the plurality of models in predicting the future behavior of the multiple actuator-sensor smart matter dynamic control system;
- increasing a weight of the at least one more successful control system model in the plurality of control system models used to predict future behavior of the multiple actuator-sensor smart matter dynamic control system relative to the weight of the at least one other model;
- and
- using the at least one more successful control system model with the increased weight to control the multiple actuator-sensor smart matter dynamic control system.
A dynamical controller of a multiple actuator-sensor smart matter dynamical control system, comprising:
- means to predict a future behavior of a multiple actuator-sensor smart matter dynamical control system using a plurality of control system models;
- means determining at least one control system model which is more successful than other models in the plurality of models in predicting future behavior of the multiple actuator-sensor smart matter dynamical control system;
- means increasing the weight of the at least one more successful control system model in the plurality of control system models used to predict future behavior of the multiple actuator-sensor smart matter dynamical control system;
- and
- means using the at least one more successful control system model to control the multiple actuator-sensor smart matter dynamical control system.
A dynamical controller of a multiple actuator-sensor smart matter dynamical control system, comprising:
- a prediction circuit usable to predict a future behavior of the multiple actuator-sensor smart matter dynamical control system using a plurality of control system models;
- a success determination circuit usable to determine at least one control system model which is more successful than at least one other model in the plurality of models in predicting the future behavior of the multiple actuator-sensor smart matter dynamical control system;
- a weight increasing circuit usable to increase the weight of the at least one more successful control system model relative to the at least one other model;
- and
- a controller that uses at least the at least one more successful control system models to control the multiple actuator-sensor smart matter dynamical control system.
Dependencies are inferred from the claim text.