Trainable drive system for a windshield wiper

• US 5,453,676 A
• Filed: 09/30/1994
• Issued: 09/26/1995
• Est. Priority Date: 09/30/1994
• Status: Expired due to Fees

• Current Assignee: ITT Automotive Electrical Systems, Inc. (ITT, Inc.)
• Original Assignee: ITT Automotive Electrical Systems, Inc. (ITT, Inc.)
• Inventors: Agrotis, Demetris A., Georgin, Marc J.
• Primary Examiner(s): Ip, Paul
• Application Number:
  US08/315,427
• Time in Patent Office: 361 Days
• Field of Search: 318/440-469, 318/560-649, 364/424.02, 395/2.68
• US Class Current: 318/643
• CPC Class Codes: B60S 1/0818 including control systems responsive to external conditions, e.g. by detection of moisture, dirt or the like
• CPC Class Codes: B60S 1/0833 Optical rain sensor
• CPC Class Codes: G06F 18/24133 Distances to prototypes
• CPC Class Codes: G06V 10/82 using neural networks
• CPC Class Codes: G06V 20/56 exterior to a vehicle by using sensors mounted on the vehicle

Abstract

A neural network provides automatic control of a windshield wiper. A rain sensor generates sensing signals indicating a rain pattern, and the neural network generates wiping demand signals indicating a wiping action desired by the driver. A training unit uses manually generated wiping supervision signals to create weight factors for the neural network.

Claims

1. An automatic drive system for a windshield wiper comprising:

(a) rain sensing means for generating sensing signals indicating rain distribution on a windshield,

(b) A neural network for using a plurality of weight factors to convert said sensing signals into wiping demand signals;

(c) a motor responsive to said wiping demand signals for driving said windshield wiper;

(d) manual adjustment means for generating wiping supervision signals; and

(e) training means connected for receiving said sensing signals and said wiping supervision signals and responsive thereto by generating said weight factors.

2. An automatic drive system for a windshield wiper comprising:

(a) rain sensing means for generating a sensing signal indicating a rain condition on a windshield;

(b) means for generating an identification signal indicating a particular human operator;

(c) demand means conditioned by training information to respond to said sensing signal and said identification signal by generating a wiping demand signal indicating a wiping action expected to be desired by said human operator for said rain condition;

(d) a motor responsive to said wiping demand signal for driving said windshield wiper; and

(e) training means connected for enabling said human operator to supply said training information to said demand means.

3. A control system for a windshield wiper comprising:

(a) rain sensing means for generating sensing signals indicating rain distribution on a windshield,

(b) network means alterable to a multiplicity of different operating states and connected for receiving said sensing signals and responding thereto by generating wiping demand signals, said wiping demand signals being adaptively correctable by changes in said operating states,

(c) manual adjustment means for generating wiping supervision signals,

(d) control means connected to a switching output terminal and responsive to signals thereon for controlling said windshield wiper,

(e) switching means connected for selective, alternative transfer of said wiping demand signals and said wiping supervision signals to said switching output terminal, and

(f) training means connected for receiving said sensing signals, said wiping demand signals and said wiping supervision signals and responsive thereto by causing corrective changes in said operating state.

4. A control system according to claim 3 wherein said rain sensing means comprises means for generating said sensing signals in a binary format, and said network means comprising node means configured for summing weight factors selected from among a group of said weight factors in accordance with binary states of said sensing signals.

5. A control system according to claim 4 wherein said training means comprises means responsive to said wiping demand signals and said wiping supervision signals for generating said weight factors.

6. A control system according to claim 5 wherein said rain sensing means comprises means for coding said sensing signals to identify a specified human operator.

7. A control system for a windshield wiper comprising:

(a) rain sensing means for generating a sensing signal indicating rain distribution on a windshield;

(b) preprocessing means for processing said sensing signal and generating a corresponding plurality of binary coded input signals;

(c) an input layer comprising a plurality of input nodes, each said input node being connected for receiving one of said binarily coded input signals and, generating parallel copies thereof;

(d) a hidden layer comprising a plurality of hidden nodes, each said hidden node having a plurality of stored weight factors and being connected for receiving one of said parallel copies from each of a plurality of said input nodes, associating one of said weight factors with each said one of said parallel copies, creating a total by summing those of said weight factors which are associated with ones of said parallel copies having a predetermined binary state, and creating a binary output signal by comparing said total against a predetermined threshold value;

(e) output means for receiving a said binary output signal from a plurality of said hidden nodes and generating a corresponding wiping demand signal;

(f) control means responsive to said wiping demand signals for controlling the operation of said windshield wiper;

(g) manual adjustment means for generating wiping supervision signals in response to manual adjustment by a human operator; and

(h) training means connected for adjusting said weight factors in response to said binarily coded input signals and said wiping supervision signals.

8. Control apparatus according to claim 7 wherein said training means comprises means for adjusting said weight factors in accordance with differences between said wiping supervision signals and said wiping demand signals.

9. Control apparatus according to claim 8 wherein said output means comprises a second said hidden layer having second hidden nodes each connected for receiving a said binary output signal from a plurality of said first mentioned hidden nodes and generating said wiping demand signal therefrom, and a layer of output nodes for relaying said wiping demand signal to said control means.

10. Method of generating a control signal for an electric motor comprising the steps of:

(1) sensing a condition to be corrected by said electric motor and generating a sensing signal indicating the severity of said condition;

(2) converting said sensing signal into a binary code comprising a plurality of binary bits;

(3) using said binary bits to set voltage states on a plurality of parallel lines;

(4) connecting each of said parallel lines to each of a plurality of hidden nodes in a neural network;

(5) supplying a weight factor to each of said hidden nodes for each said parallel lines connected thereto;

(6) supplying a bias weight to each of said hidden nodes;

(7) causing each of said hidden nodes to perform a weighted multiplication by calculating a sum for its said bias weight and those of its said weight factors associated with a said parallel line which is set to a predetermined voltage state;

(8) causing each said hidden node to generate a binary output bit by comparing its said sum against a predetermined threshold value; and

(9) generating said control signal by combining binary output bits from said hidden nodes.