Spiking Neuron PID Controller with Expansion Kernels

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing robotic control systems using spiking neuron networks face challenges in effectively encoding and processing sensory data of varying dynamical range and nature, requiring cumbersome sensor model changes and struggling with mixed or degraded sensor inputs.

Innovation Solution

A proportional-integral-derivative controller apparatus utilizing a spiking neuron network with an expansion kernel, comprising receptive fields that encode inputs into spiking outputs through proportional, integrating, and differentiating operators, and a reinforcement learning process to adjust connection efficacy based on external signals.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If spiking neuron networks are used to control robotic systems, then the system can process sensory data through biological-inspired mechanisms, but the system struggles to effectively encode and process sensory data of varying dynamical range and nature

Engineering Contradiction:

Improveability to process varying sensory dataVSAvoidcomplexity of sensor model changes

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent implements a universal encoding mechanism using spiking neurons with adjustable receptive fields that can handle multiple types of sensory inputs (analog, digital, mixed) through a single unified framework. The expansion kernel with configurable neurons serves as a multi-functional interface that adapts to different sensor types without requiring separate processing pathways, thereby resolving the contradiction between versatility and complexity.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system employs dynamic adjustment of connection efficacies between spiking neurons and sensory inputs through reinforcement learning. The receptive fields and encoding parameters are not fixed but can be adapted in real-time based on the nature and dynamical range of the sensory data, allowing the system to maintain effectiveness across varying sensor conditions without structural changes.

Inventive Principle:

Principle #15Dynamics

2Reliability

If traditional sensor models are used in robotic control systems, then the system can process specific types of sensor inputs, but the system requires cumbersome model changes when sensors are upgraded or replaced

Engineering Contradiction:

Improveconsistency of control processingVSAvoidease of sensor upgrade and replacement

Core Design Contradiction:

ReliabilityVSEase of manufacture

Solution Approach 1:

The spiking neuron expansion kernel serves as a universal interface that can accommodate different sensor types through configuration rather than structural change. The same neural network architecture processes both analog and digital sensor inputs by adjusting receptive field parameters, eliminating the need to redesign control models when sensors are upgraded or replaced.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system maintains reliability during sensor changes by adjusting parameters of the spiking neuron network (such as receptive field characteristics and connection efficacies) rather than changing the fundamental processing architecture. This parameter-based adaptation allows seamless integration of new sensors while maintaining consistent control processing behavior.

Inventive Principle:

Principle #35Parameter changes

3Adaptability or versatility

If spiking neuron networks process mixed or degraded sensor inputs, then the system can operate in real-world conditions, but the system struggles to maintain performance with mixed or degraded inputs

Engineering Contradiction:

Improveability to handle mixed sensor inputsVSAvoidperformance consistency with degraded inputs

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The system employs reinforcement learning as a feedback mechanism that continuously monitors the quality and nature of sensor inputs and adjusts the spiking neuron network's connection efficacies accordingly. This feedback loop enables the system to adapt to mixed or degraded inputs by learning optimal encoding strategies from actual performance outcomes, thereby maintaining reliability despite input variations.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The spiking neuron network dynamically adjusts its receptive fields and connection weights in response to the characteristics of incoming sensor data. When inputs are mixed or degraded, the system's parameters evolve in real-time to optimize processing, allowing it to maintain performance consistency across varying input conditions through adaptive rather than static processing.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS9082079B1Proportional-integral-derivative controller effecting expansion kernels comprising a plurality of spiking neurons associated with a plurality of receptive fields

Publication Date: 2015.07.14 BRAIN CORP

AI summary

Adaptive proportional-integral-derivative controller apparatus of a plant may be implemented. The controller may comprise an encoder block utilizing basis function kernel expansion technique to encode an arbitrary combination of inputs into spike output. The basis function kernel may comprise one or more operators configured to manipulate basis components. The controller may comprise spiking neuron network operable according to reinforcement learning process. The network may receive the encoder output via a plurality of plastic connections. The process may be configured to adaptively modify connection weights in order to maximize process performance, associated with a target outcome. Features of the input may be identified and used for enabling the controlled plant to achieve the target outcome.