Mobile Body Control Using Hardware Dynamic Identification Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In automatic driving systems, it is challenging to detect and address malfunctions in hardware units such as engines and brakes, especially when the driver's intervention is limited, as the difference between target and actual outputs cannot be accurately compensated without knowing the cause, and verifying the hardware unit during operation is difficult.
Innovation Solution
A control device for mobile bodies that includes identification models to simulate hardware unit dynamics, allowing for the specification of malfunctions and adjustment of control parameters to match normative output values, thereby minimizing operational disruptions. This device uses hard and soft models to calculate and update dynamic characteristics, enabling appropriate control without affecting the vehicle's operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If feedback control is used to compensate for the difference between actual and target output, then control accuracy can be improved, but if the cause of the difference is unknown, appropriate compensation cannot be performed
Solution Approach 1:
The patent introduces identification models as intermediary systems that simulate hardware unit behavior. These models act as mediators between the control section and actual hardware, allowing the system to infer malfunction causes by comparing model predictions with actual outputs without directly interfering with hardware operation. The models provide the missing information about malfunction causes that feedback control alone cannot obtain.
Solution Approach 2:
The patent creates virtual copies of the hardware unit through identification models (hard models and soft models). These copies replicate the dynamic characteristics and behavior of the actual hardware, allowing the system to analyze potential malfunctions and their causes in a virtual environment before applying corrections to the actual system, thus preserving operational safety while enabling accurate diagnosis.
2Difficulty of detecting and measuring
If verification of the hardware unit is performed by changing control content, then malfunction cause can be identified, but the mobile body operation cannot be verified while in operation
Solution Approach 1:
The patent uses identification models as virtual copies to perform verification operations. Instead of physically testing the hardware by changing control content during operation, the system applies control variations to the virtual models and compares their outputs with actual hardware responses. This allows comprehensive malfunction detection without disrupting actual vehicle operation.
Solution Approach 2:
The patent performs preliminary verification by updating the identification models with actual hardware output data before actual malfunctions occur. The adjustment section continuously refines the models' accuracy by comparing predicted versus actual outputs, so that when malfunctions do occur, the models are already calibrated to accurately detect and diagnose the specific type of malfunction.
3Reliability
If the driver's determination based on experience is used to cope with malfunction, then appropriate response can be achieved, but this is difficult in automatic driving where driver operation is limited
Solution Approach 1:
The patent enables the control system to diagnose and respond to malfunctions autonomously without driver intervention. The identification models automatically detect malfunction causes by comparing their predictions with actual hardware outputs, and the control section automatically adjusts control content based on this diagnosis. This self-diagnosing, self-correcting capability replaces the driver's experience-based judgment with automated intelligent control.
Solution Approach 2:
The patent replaces the driver's mechanical decision-making process with an automated identification and control system. Instead of relying on the driver's sensory perception and experience-based judgment, the system uses mathematical models and computational algorithms to detect malfunctions and determine appropriate responses, thereby maintaining reliable malfunction handling while enabling full automation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A controller 10 with a hardware unit 1 mounted therein includes a control section 100 configured to actually control the hardware unit 1, an identification model 120 including a hard model 121 obtained by modeling a dynamic characteristic of the hardware unit 1 and a soft model 122 configured to execute same processing as processing performed on the hard model 121 by the control section 100, and an adjustment section 142 configured to update a dynamic characteristic of a model of the hardware unit 1 in the hard model 121 such that an output value of the hard model 121 obtained by processing of the soft model 122 matches an actual output value of the hardware unit 1.