Hardwired Processor Core for Real-Time RBF and MLP Calculation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current computing capacities struggle to calculate data-based functional models, such as multilayer perceptron and RBF models, within real-time requirements for control units in motor vehicles due to high computing demands, necessitating a hardware-based solution for efficient model calculation.

Innovation Solution

A model calculating unit with a hardwired processor core designed to selectively calculate RBF or multilayer perceptron models, utilizing coupled functional blocks, shared memory areas, and a state machine to perform calculations efficiently, allowing for variable neuron numbers and activation functions, and supporting partial derivations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If data-based functional models (Gaussian process model, multilayer perceptron model) are used to achieve sufficient modeling precision, then modeling precision is improved, but computing capacity requirements increase significantly

Engineering Contradiction:

Improvemodeling precisionVSAvoidcomputing capacity

Core Design Contradiction:

Measurement precisionVSPower

Solution Approach 1:

The patent segments the computational workload by separating the model calculating unit from the microprocessor. The model calculating unit is dedicated to executing specific model calculations (RBF, MLP, Gaussian process) while the microprocessor handles other control tasks. This segmentation allows the system to achieve high modeling precision through specialized hardware without overloading the general-purpose microprocessor, thus resolving the contradiction between modeling precision and computing capacity requirements.

Inventive Principle:

Principle #1Segmentation

2Measurement precision

If a great number of supporting points are used in data-based functional models, then modeling precision is improved, but the computing capacity required for real-time calculation increases

Engineering Contradiction:

Improvemodeling precisionVSAvoidreal-time calculation capability

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The patent replaces the software-based mechanical calculation system with a hardware-based model calculating unit. This hardware unit is specifically designed to perform the computationally intensive operations required for data-based functional models with many supporting points. By substituting software execution with dedicated hardware circuitry, the system achieves both high modeling precision (through sufficient supporting points) and real-time calculation capability (through parallel hardware processing).

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The model calculating unit acts as an intermediary between the microprocessor and the complex mathematical models. It receives model parameters and input data from the microprocessor, performs the computationally intensive calculations using dedicated hardware resources, and returns results to the microprocessor. This intermediary structure enables the system to handle complex models with many supporting points in real-time without burdening the main microprocessor.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Productivity

If hardware-based model calculating units are provided to enable real-time calculation, then real-time calculation capability is improved, but device complexity increases

Engineering Contradiction:

Improvereal-time calculation capabilityVSAvoiddevice complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The model calculating unit is designed with multi-functionality to handle multiple types of data-based functional models (RBF models, MLP models, Gaussian process models) within a single hardware device. This universal design allows the system to achieve real-time calculation capability for various complex models without requiring separate dedicated hardware for each model type, thus improving productivity while limiting the increase in device complexity through resource sharing and consolidation.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS11537090B2Model calculating unit and control unit for selectively calculating an RBF model, a gaussian process model and an MLP model

Publication Date: 2022.12.27 ROBERT BOSCH GMBH

AI summary

A model calculating unit for the selective calculation of an RBF model or of a neural layer of a multilayer perceptron model having a hardwired processor core designed in hardware for calculating a fixedly specified computing algorithm in coupled function blocks. The processor core is designed to calculate an output variable for an RBF model as a function of one or multiple input variables of an input variable vector, of supporting points, of length scales, of parameters specified for each supporting point, the processor core furthermore being designed to calculate an output variable for each neuron for the neural layer of the multilayer perceptron model having a number of neurons as a function of the one or the multiple input variables of the input variable vector, of a weighting matrix having weighting factors and an offset value specified for each neuron.