CNN programamble topographic sensory device

• US 5,717,834 A
• Filed: 01/11/1996
• Issued: 02/10/1998
• Est. Priority Date: 08/26/1993
• Status: Expired due to Term

• Current Assignee: Frank S. Werblin, Leon O. Chua, Tamas Roska
• Original Assignee: Frank S. Werblin, Leon O. Chua, Tamas Roska
• Inventors: Chua, Leon O., Werblin, Frank S., Roska, Tamas
• Primary Examiner(s): Hafiz, Tariq R.
• Application Number:
  US08/583,897
• Time in Patent Office: 761 Days
• Field of Search: 382/155-159, 395/20-25, 395/27
• US Class Current: 706/16
• CPC Class Codes: G06N 3/045 Combinations of networks
• CPC Class Codes: G06N 3/063 using electronic means

Abstract

The main design components underlying the implementation of physiologically faithful retina and other topographic sensory organ models on Cellular Neural Network (CNN) universal chips is discussed. If the various retinas are implemented on a CNN universal chip, in a programmable way, it can be called a "CNN bionic eye", a device capable of performing a broad range of image processing functions similar to those performed by biological retinas. The CNN universal machine has the special properties that it is 1) programmable and 2) includes local memory. Programming is stored in analog and logical form (the analogic program) generated by an analogic programming and control unit, so the functions of the CNN universal machine can be modified as a function of complex internal and external constraints. Further, several CNN bionic eyes and other topographic sensory modalities can be combined on a single CNN universal chip, and, for more complex sensory tasks, the necessary physical microsensors to provide the input signals can be implemented on the chip, in most instances.

Claims

1. A method for creating a model of functions of a plurality of distinct topographic physiological sensory organs in response to a set of stimuli, with each of said functions performed by an array of cells arranged in a pattern having a grain density, with said grain density being defined as the number of cells per square unit said method comprising the steps of:

a. selecting a plurality of topographic physiological sensory organs to be modelled by said model as a part of the same process;

b. selecting stimuli to which said model is to respond from a set of stimuli to which said organs selected in step a. respond;

c. selecting the grain density and shape for each of said arrays of cells of said organs selected in step a. appropriate to perform said functions;

d. configuring a Cellular Neural Network (CNN) universal machine to have an array of cells that has a shape and grain density corresponding to the largest shape and greatest grain density, respectively, of said organs selected in step a.;

e. selecting a plurality of sets of sensors with one set of sensors to detect each of said stimuli selected in step b.;

f. interfacing at least one sensor of one of said plurality of sets of sensors selected in step e. to each cell of said CNN universal machine as configured in step d. and matching the grain density and shape for the corresponding organ as selected in step b., with each of said sets of sensors being responsive to one of said stimuli selected in step b. and individually developing an output signal compatible with that cell of said CNN universal machine to which said sensor is interfaced, with said individual signal being representative of a corresponding stimuli received by said individual sensor; and

g. programming said CNN universal machine to implement said functions of said topographic sensory organs as part of the same process.

2. A method as in claim 1 wherein:

step a. includes selecting the same topographic physiological sensory organ from two different organisms for inclusion in the same said model.

3. A method as in claim 1 wherein:

step a. includes selecting two different topographic physiological sensory organs from the same organism for inclusion in the same said model; and

step b. includes selecting said set of stimuli to which said model is to respond from a combination of stimuli to which each of said organs selected in step a. respond.

4. A method as in claim 1 wherein in step g. said functions of said topographic sensory organs are implemented sequentially.

5. A method as in claim 1 further includes the step of:

h. correlating the responses of said model to said set of stimuli resulting from a common event.

6. A method for creating a model of a function of a topographic physiological sensory organ in response to a set of stimuli, with said function performed by an array of cells arranged in a pattern having a grain density, with said grain density being defined as the number of cells per square unit, said method comprising the steps of:

a. selecting a topographic physiological sensory organ to be modelled by said model;

b. selecting the stimuli to which said model is to respond from a set of stimuli to which said organ selected in step a. responds;

c. selecting the grain density and shape said array of cells of said organ selected in step a. appropriate to perform said function;

d. configuring a Cellular Neural Network (CNN) universal machine to have an array of cells that has said shape and said grain density selected in step c. of said organ selected in step a.;

e. selecting a set of sensors to detect said stimuli selected in step b.;

f. interfacing one of said set of sensors selected in step e. to each cell of said CNN universal machine as configured in step d., with each of said set of sensors being responsive to said stimuli selected in step b. and individually developing an output signal compatible with that cell of said CNN universal machine to which said sensor is interfaced, with said individual signal being representative of said stimuli received by said individual sensor;

g. programming said CNN universal machine to implement said function of said topographic sensory organ, including

programming a defect in the normal performance of said selected physiologically topographic sensory organ being modelled; and

h. performing diagnostic tests following step g. to research potential corrective actions that may be possible to correct or improve the performance of said selected physiologically topographic sensory organ in a living organism.

7. A method for creating a model of a function of two topographic physiological sensory organs in response to a stimuli, with said function performed by an array of cells arranged in a pattern having a grain density, with said grain density being defined as the number of cells per square unit, said method comprising the steps of:

a. selecting the same topographic physiological sensory organ to be modelled by said model from two different organisms;

b. selecting a set of stimuli to which said model is to respond from a combination of stimuli to which each of said organs selected in step a. respond;

c. selecting a grain density and shape for said array of cells of said organ selected in step a. appropriate to perform said function, said grain density and shape corresponding to the greatest grain density and largest shape, respectively, of said two organs selected in step a.

d. configuring a Cellular Neural Network (CNN) universal machine to have an array of cells that has said shape and said grain density selected in step c. of said organ selected in step a.;

e. selecting a set of sensors to detect said set of stimuli selected in step b.;

f. interfacing one of said set of sensors selected in step e. to each cell of said CNN universal machine as configured in step d. with each of said set of sensors being in proportion to said stimuli selected in step b. and individually developing an output signal compatible with that cell of said CNN universal machine to which said sensor is interfaced, with said individual signal being representative of said stimuli received by said individual sensor;

g. programming said CNN universal machine to implement said function of said topographic sensory organ, including programming a defect in the normal performance of said selected physiologically topographic sensory organs, or the interaction between them, being modelled; and

h. performing diagnostic tests following step g. to research potential corrective actions that may be possible to correct or improve the performance of said selected physiologically topographic sensory organs and the interaction between them.

8. Apparatus to create a model of a function of a topographic physiological sensory organ having an array of cells arranged in a pattern size and shape, and grain density, with said grain density being defined as the number of cells per square unit, with said physiological sensory organ performing in response to a set of externally applied stimuli, said apparatus comprising:

a Cellular Neural Network (CNN) universal machine having an array of CNN cells in said pattern size, shape and grain density of said physiological sensory organ;

a set of sensors with one of said sensors interfacing with each CNN cell, each of said sensors disposed to be individually responsive to said externally applied stimuli and to individually develop a signal responsive to said stimuli received by said individual sensor and then to apply said individually developed signal to the corresponding one of said CNN cells;

a processor connected to said CNN universal machine to sequentially program said CNN universal machine to simulate the performance of a second series of steps in a selected order in response to signals from said set of sensors in response to said externally applied stimuli, said steps of said second series selected from said first series of steps performed by said topographic physiological sensory organ; and

a memory interfaced with said processor;

wherein said processor stores to said memory the results of each step as necessary to perform subsequent steps; and

wherein said processor programs said CNN cells to simulate performance of a selected second series of steps with those steps selected from said first series of steps of two different topographic sensory organs in the same organism.

9. Apparatus to create a model of a function of a topographic physiological sensory organ having an array of cells arranged in a pattern size and shape, and grain density, with said grain density being defined as the number of cells per square unit, with said physiological sensory organ performing in response to a set of externally applied stimuli, said apparatus comprising:

a Cellular Neural Network (CNN) universal machine having an array of CNN cells in said pattern size, shape and grain density of said physiological sensory organ;

a set of sensors with one of said sensors interfacing with each CNN cell, each of said sensors disposed to be individually responsive to said externally applied stimuli and to individually develop a signal responsive to said stimuli received by said individual sensor and then to apply said individually developed signal to the corresponding one of said CNN cells;

a processor connected to said CNN universal machine to sequentially program said CNN universal machine to simulate the performance of a second series of steps in a selected order in response to signals from said set of sensors in response to said externally applied stimuli, said steps of said second series selected from said first series of steps performed by said topographic physiological sensory organ; and

a memory interfaced with said processor;

wherein said processor stores to said memory the results of each step as necessary to perform subsequent steps; and

wherein said processor programs said CNN cells in said second series of steps to faithfully simulate said first series of steps and the sequence performed by a selected physiologically topographic sensory organ; and

said processor stores in said memory program variations for use by said processor to alter the program of said CNN cells to simulate a defect in the normal performance of said selected physiologically topographic sensory organ being modelled to permit the performance of diagnostic tests on said model to research potential corrective actions that may be possible to correct or improve the performance of said selected physiologically topographic sensory organ in a living organism.

10. Apparatus to create a model of a function of a topographic physiological sensory organ having an array of cells arranged in a pattern size and shape, and grain density, with said grain density being defined as the number of cells per square unit, with said physiological sensory organ performing in response to a set of externally applied stimuli, said apparatus comprising:

a Cellular Neural Network (CNN) universal machine having an array of CNN cells in said pattern size, shape and grain density of said physiological sensory organ;

a set of sensors with one of said sensors interfacing with each CNN cell, each of said sensors disposed to be individually responsive to said externally applied stimuli and to individually develop a signal responsive to said stimuli received by said individual sensor and then to apply said individually developed signal to the corresponding one of said CNN cells;

a processor connected to said CNN universal machine to sequentially program said CNN universal machine to simulate the performance of a second series of steps in a selected order in response to signals from said set of sensors in response to said externally applied stimuli, said steps of said second series selected from said first series of steps performed by said topographic physiological sensory organ; and

a memory interfaced with said processor;

wherein said processor stores to said memory the results of each step as necessary to perform subsequent steps; and

wherein said processor programs said CNN cells in said second series of steps to faithfully simulate said first series of steps and sequence of two selected topographic physiological sensory organs within the same organism and interactions therebetween; and

said memory stores program variations for use by said processor to alter the program on said CNN cells to simulate a defect in the normal performance and interaction between said two selected physiologically topographic sensory organs to permit the performance of diagnostic tests on said model to research potential corrective actions that may be possible to correct or improve the performance of said two topographic physiological sensory organs and the interaction between them in a single living organism.

11. Apparatus to create a model of a function of a topographic physiological sensory organ having an array of cells arranged in a pattern size and shape, and grain density, with said grain density being defined as the number of cells per square unit, with said physiological sensory organ performing in response to a set of externally applied stimuli, said apparatus comprising:

a Cellular Neural Network (CNN) universal machine having an array of CNN cells in said pattern size, shape and grain density of said physiological sensory organ;

a set of sensors with one of said sensors interfacing with each CNN cell, each of said sensors disposed to be individually responsive to said externally applied stimuli and to individually develop a signal responsive to said stimuli received by said individual sensor and then to apply said individually developed signal to the corresponding one of said CNN cells;

a processor connected to said CNN universal machine to sequentially program said CNN universal machine to simulate the performance of a second series of steps in a selected order in response to signals from said set of sensors in response to said externally applied stimuli, said steps of said second series selected from said first series of steps performed by said topographic physiological sensory organ; and

a memory interfaced with said processor;

wherein said processor stores to said memory the results of each step as necessary to perform subsequent steps; and

wherein said processor programs said CNN cells to simulate performance of a selected second series of steps, with those steps selected from said first series of steps of the same topographic sensory organ of two different living organisms.

12. A sensor to detect a selected applied topographic stimulus that is outside the usual range of human physiological sensory organs and to present a representation of said stimulus in a form that is within the usual stimulus range of a selected human physiological sensory organ, said sensor comprising:

a Cellular Neural Network (CNN) universal machine having an array of CNN cells of a selected pattern size and shape, and grain size density to provide a desired resolution of said topographic stimulus, wherein said grain density is the number of cells per square unit;

a set of sensors for measuring stimulus outside the usual range of human physiological sensory organs, with one of said sensors interfacing individually with each CNN cell, each of said sensors disposed to be individually responsive to said topographic stimulus and to individually develop a signal in response to said stimulus received by said individual sensor, and then to apply said individually developed signal to the corresponding one of said CNN cells;

an output device capable of presenting a response in a form that is within the usual stimulus range of a selected human physiological sensory organ; and

a processor connected to said CNN universal machine and to said output device to sequentially program said CNN universal machine and to process signals from each of said CNN cells to format a representation of said stimulus to be applied to said output device.

13. A method for sensing a selected applied topographic stimulus that is outside the usual range of human physiological sensory organs and to present a representation of said stimulus to an output device to present a response in a form that is within the usual stimulus range of a selected human physiological sensory organ, said method comprising the steps of:

a. configuring a Cellular Neural Network (CNN) universal machine to have an array of cells of a selected pattern size and shape, and grain density to provide the desired resolution of said topological stimulus, wherein said grain density is the number of cells per square unit;

b. selecting a set of sensors for measuring stimulus outside the usual range of human physiological sensory organs to detect said applied topographic stimulus;

c. interfacing one of said set of sensors selected in step b. to each cell of said CNN universal machine as configured in step a. with each of said set of sensors being responsive to said applied stimulus and individually developing an output signal compatible with the cell of said CNN universal machine to which it is interfaced with said individual signal being in proportion to said stimulus received by said individual sensor; and

d. programming said CNN universal machine to process signals from each of said CNN cells and to format a representation of said stimulus to be applied to said output device to present a response in a form that is within the usual stimulus range of a selected human physiological sensory organ.

14. A method as in claim 13 wherein said method adaptively models sensory function that is outside the usual range of human physiological sensory organs.

15. A method as in claim 13 wherein said modelled sensory function exceeds the complexity and speed of operation of human physiological sensory organs.

16. A method for sensing a selected applied topographic stimulus that is outside the usual range of vertebrate physiological sensory organs and to present a representation of said stimulus to an output device to present a response in a form that is within the usual stimulus range of a selected vertebrate physiological sensory organ, said method comprising the steps of:

a. configuring a Cellular Neural Network (CNN) universal machine to have an array of cells of a selected pattern size and shape, and grain density to provide the desired resolution of said topological stimulus, wherein said grain density is the number of cells per square unit;

b. selecting a set of sensors for measuring stimulus outside the usual range of vertebrate physiological sensory organs to detect said applied topographic stimulus;

c. interfacing one of said set of sensors selected in step b. to each cell of said CNN universal machine as configured in step a. with each of said set of sensors being responsive to said applied stimulus and individually developing an output signal compatible with the cell of said CNN universal machine to which it is interfaced with said individual signal being in proportion to said stimulus received by said individual sensor; and

d. programming said CNN universal machine to process signals from each of said CNN cells and to format a representation of said stimulus to be applied to said output device to present a response in a form that is within the usual stimulus range of a selected vertebrate physiological sensory organ.

17. A method as in claim 16 wherein said method adaptively models sensory function that is outside the range of vertebrate physiological sensory organs.

18. A method as in claim 16 wherein said modelled sensory function exceeds the complexity and speed of operation of vertebrate physiological sensory organs.