Prior art

The United States Patent (US) 12163947B2, titled "Method and system for characterizing undebarked wooden logs and computing optimal debarking parameters in real time," was granted on December 10, 2024, and has an adjusted expiration date of May 26, 2043. The patent's priority date is August 27, 2020. The invention focuses on using deep learning artificial intelligence (AI) models for contactless characterization of undebarked logs upstream from a debarking device to adjust debarking parameters in real time, thereby minimizing fiber loss and residual bark.

Based on the "Patent Citations" section of US12163947B2 and the detailed descriptions provided within its "BACKGROUND OF THE INVENTION," the following are the most relevant prior art references:

Most Relevant Prior Art for US12163947B2

1. U.S. Patent No. 9,588,098 B2

• Full Citation: US9588098B2, "Optical method and apparatus for identifying wood species of a raw wooden log" by Centre De Recherche Industrielle Du Quebec.
• Publication/Filing Date: Priority date: 2015-03-18; Publication date: 2017-03-07.
• Brief Description: This patent discloses an optical method for identifying wood species by subdividing images into small squares, calculating texture statistics (Local Binary Patterns (LBPs) and histograms), and classifying these statistics using a simple neural network, support vector machine (SVM), multivariate linear model, or static gain matrix. The classification provides probable species indications for image regions.
• Potential Anticipation (35 U.S.C. § 102): While this patent identifies characteristics (species) of raw wooden logs using an optical method and a form of neural network, US12163947B2 explicitly distinguishes it. US12163947B2 states that such "classic texture methods are not as accurate as newer deep learning AI techniques," noting that the "small local texture images" used by US9588098B2 "may not all contain special characteristics of a species." Furthermore, US9588098B2 does not teach the subsequent steps of computing optimal debarking parameters based on identified characteristics (including intensity levels of knots, moisture, etc.) and sending these parameters to a debarker for real-time adjustment. Therefore, it potentially anticipates the broad concept of "identifying characteristics... using a trained model" (Claim 1) or "identifying attributes... using the deep learning model" (Claim 11) in a very general sense, but it does not anticipate the use of a deep learning model for broad log characterization, nor the real-time debarker optimization loop as claimed in US12163947B2.

2. U.S. Patent No. 10,099,400 B2 (and related US20130333805A1, CA2780202A1)

• Full Citation: US10099400B2, "Method and System for Detecting the Quality of Debarking at the surface of a Wooden Log" by Centre De Recherche Industrielle Du Québec.
• Publication/Filing Date: Priority date: 2012-06-19; Publication date: 2018-10-16.
• Brief Description: This patent discloses a system for measuring the efficiency of debarking downstream of the debarker. It provides data that can be used to adjust the debarker, either by human intervention or an automated process. US12163947B2 notes that such a downstream system "presents hints on what should have been performed during the debarking process" but "cannot predict sudden unexpected changes in the incoming undebarked logs." It is suitable only for "steady lines of production having a low variation in log characteristics."
• Potential Anticipation (35 U.S.C. § 102): This patent addresses debarking quality and adjustment, but critically, it operates downstream of the debarker. This fundamentally distinguishes it from Claim 1 and Claim 11 of US12163947B2, which involve upstream, real-time characterization of undebarked logs to predictively compute and send optimal parameters before debarking the specific log. The lack of upstream, predictive capability for incoming logs and the absence of a deep learning model for undebarked log characterization mean it does not anticipate the core novelty of US12163947B2.

3. U.S. Patent No. 8,215,347 B2

• Full Citation: US8215347B2, "Apparatus and methods for controlled debarking of wood" by Fpinnovations.
• Publication/Filing Date: Priority date: 2008-10-03; Publication date: 2012-07-10.
• Brief Description: This patent describes an apparatus and method comprising a mechanical surface scraper used to determine optimal operating parameters of a debarker. US12163947B2 criticizes this apparatus as "hard to implement, having reliability and ruggedness issues" due to its operation in a harsh environment.
• Potential Anticipation (35 U.S.C. § 102): This patent is relevant for determining debarker operating parameters. However, its use of a "mechanical surface scraper" means it employs a contact-based method, which directly contrasts with the "contactless characterization" element specified in Claim 1 and Claim 11 of US12163947B2. Furthermore, it does not mention the use of deep learning models for log characterization. Thus, it does not anticipate Claim 1 or Claim 11.

4. U.S. Patent No. 6,526,154 B1

• Full Citation: US6526154B1, "Method and apparatus for determining the portion of wood material present in a stream of bark" by Andritz-Patentverwaltungs-Gmbh.
• Publication/Filing Date: Priority date: 1997-05-19; Publication date: 2003-02-25.
• Brief Description: This invention is described as "useful for determining debarking quality." However, US12163947B2 explicitly states it is "useless for adjusting the debarker in real time when the wood species or moisture content of incoming logs varies."
• Potential Anticipation (35 U.S.C. § 102): While related to debarking quality, this patent is explicitly dismissed by US12163947B2 as incapable of "real-time adjustment" based on varying log characteristics. This directly indicates it does not teach the real-time computation and sending of operating parameters to the debarker based on identified characteristics, which is a key component of Claim 1 and Claim 11.

5. WO2018169712A1

• Full Citation: WO2018169712A1, "Method of board lumber grading using deep learning techniques" by Lucidyne Technologies, Inc.
• Publication/Filing Date: Priority date: 2017-03-13; Publication date: 2018-09-20.
• Brief Description: While not detailed in the background, its title indicates the use of "deep learning techniques" for "board lumber grading."
• Potential Anticipation (35 U.S.C. § 102): This patent is relevant for applying deep learning to wood processing. However, it applies deep learning to "board lumber grading," which pertains to processed wood, not the upstream characterization of undebarked logs for debarking optimization. The application and purpose differ significantly from the claims of US12163947B2.

In summary, US12163947B2 carefully distinguishes itself from the cited prior art by highlighting its unique combination of contactless, upstream characterization of undebarked logs using deep learning models to compute and send optimal debarking parameters in real time for the immediate adjustment of the debarker, considering various log characteristics and their intensity levels. None of the listed prior art, as described within US12163947B2, appears to fully anticipate all these combined elements.