Obviousness

The US patent 12163947 describes a method and system for characterizing undebarked wooden logs in real time using deep learning AI models, upstream from a debarking device, to compute and send optimal debarking parameters. This aims to minimize fiber loss and residual bark.

An analysis under 35 U.S.C. § 103 for obviousness requires identifying combinations of prior art that would have made the claimed invention obvious to a person having ordinary skill in the art (POSITA) at the time of the invention (priority date: 2020-08-27).

Key Inventive Concepts of US12163947 (Independent Claims 1 and 11):

• Upstream, Contactless Characterization: Measuring attributes of an undebarked log before it reaches the debarker.
• Deep Learning Model: Identifying log characteristics (species, moisture, knots) using a trained deep learning model.
• Real-time Optimization: Computing and sending optimal debarking parameters to the debarker based on these identified characteristics in real time.
• Parameter Table/Indexing: Retrieving operating parameters from a table associated with log characteristics, potentially considering intensity levels.

Prior Art References and their Relevance:

1. U.S. Pat. No. 9,588,098 B2 (CRIQ, published 2017-03-07): Discloses an optical method for identifying wood species of "raw wooden logs" by subdividing images, calculating texture statistics (Local Binary Patterns - LBPs, histogram), and classifying them using a "simple neural network" or a support vector machine (SVM). This patent clearly teaches upstream, contactless, image-based characterization of logs to determine intrinsic characteristics, specifically wood species [cite: The U.S. Pat. No. 9,588,098 B2 discloses an optical method for subdividing images into a plurality of small squares for which a plurality of texture statistics is calculated. The U.S. Pat. No. 9,588,098 B2 calculates Local Binary Patterns (LBPs) and histogram and performs statistical analysis and classification based on LBPs and histogram. The classification processing is carried out for the calculated vectors associated with all image regions, resulting in a set of probable species indications.].
2. WO2018/169712 A1 (Lucidyne Technologies, Inc., published 2018-09-20): Teaches a "Method of board lumber grading using deep learning techniques" [cite: WO2018169712A1 discloses a method of board lumber grading using deep learning techniques.]. This reference demonstrates the application of deep learning algorithms specifically for analyzing wood characteristics from images, albeit for graded lumber rather than undebarked logs.
3. U.S. Pat. No. 10,099,400 B2 (CRIQ, published 2018-10-16): Titled "Method and System for Detecting the Quality of Debarking at the surface of a Wooden Log," this patent discloses measuring debarking efficiency downstream of the debarker and using that data to adjust the debarker either by human intervention or an automated process [cite: a system such as the one disclosed in the U.S. Pat. No. 10,099,400 B2 titled “Method and System for Detecting the Quality of Debarking at the surface of a Wooden Log” discloses measuring the efficiency of debarking downstream of the debarker.]. The background of US12163947 explicitly discusses the limitations of such downstream feedback systems, noting they "cannot predict sudden unexpected changes in the incoming undebarked logs" [cite: The main drawback of the prior art feedback systems, however, is that they cannot predict sudden unexpected changes in the incoming undebarked logs.].
4. General Knowledge in the Art (as reflected in US12163947's background): The patent itself acknowledges that "Debarking process optimisation is a very complex task. It requires detailed knowledge of the incoming material and knowledge of the control of the debarking devices in order to decrease fiber loss and remaining bark quantity" [cite: Debarking process optimisation is a very complex task. It requires detailed knowledge of the incoming material and knowledge of the control of the debarking devices in order to decrease fiber loss and remaining bark quantity.]. It also states that "Debarking parameters such as rotational speed and tools pressure are related to the intrinsic characteristics of the logs" [cite: Debarking parameters such as rotational speed and tools pressure are related to the intrinsic characteristics of the logs.]. Furthermore, the patent notes that "Such classic texture methods [like those of '098] are not as accurate as newer deep learning AI techniques" [cite: Such classic texture methods are not as accurate as newer deep learning AI techniques.].

Obviousness Combination: US 9,588,098 B2 + WO2018/169712 A1 + US 10,099,400 B2

A person having ordinary skill in the art (POSITA) in sawmill automation and debarking would have been motivated to combine these prior art references to arrive at the claimed invention.

Motivation for Combination:

1. Improving Log Characterization Accuracy (US 9,588,098 B2 + WO2018/169712 A1):
   A POSITA would have recognized the limitations of the "classic texture methods" and "simple neural network" used in US 9,588,098 B2 for characterizing raw logs, as explicitly noted in the background of US12163947. Given the advancements in artificial intelligence, particularly the rise of deep learning, and its proven superiority for image analysis and classification, a POSITA would have been motivated to replace the older classification techniques of '098 with "deep learning techniques." The WO2018/169712 A1 patent provides a clear example of applying deep learning specifically to wood analysis for grading. Therefore, it would have been obvious to a POSITA to enhance the upstream optical log characterization of '098 by incorporating deep learning models for more accurate and robust identification of characteristics like species, moisture, and knot presence.

2. Transitioning from Reactive to Predictive Debarker Control (US 9,588,098 B2 / WO2018/169712 A1 + US 10,099,400 B2 + General Knowledge):
   The industry faced a known problem of optimizing debarking efficiency, as highlighted by US 10,099,400 B2, which attempted to address it with a downstream feedback system. However, US12163947's background acknowledges the critical drawback of such systems: their inability to "predict sudden unexpected changes in the incoming undebarked logs" [cite: The main drawback of the prior art feedback systems, however, is that they cannot predict sudden unexpected changes in the incoming undebarked logs.]. Simultaneously, it was well-known that "Debarking parameters such as rotational speed and tools pressure are related to the intrinsic characteristics of the logs" [cite: Debarking parameters such as rotational speed and tools pressure are related to the intrinsic characteristics of the logs.].

   A POSITA, seeking to overcome the limitations of reactive debarker control and knowing the relationship between log characteristics and optimal debarking parameters, would be strongly motivated to leverage the improved upstream characterization made possible by combining '098 and WO'712. Once accurate log characteristics (e.g., species, moisture level, knot intensity, which are explicitly mentioned as desired characteristics to identify for debarking optimization in the background) could be determined in real time before debarking, it would be a straightforward engineering step to use this predictive information to compute and send optimal debarking parameters to the debarker. The concept of "computing operating parameters of a debarker based on the identified characteristics" (Claim 1) and "sending the computed operating parameters to the debarker" (Claim 1) is a logical extension of known control principles, moving the feedback loop further upstream for proactive adjustment.

3. Implementation of Control Parameters (Claims 5, 6, 11):
   The idea of having "operating parameters indexed in a table having operating parameters associated with one or more log characteristics" (Claim 5) and taking into account "intensity level" (Claim 6) is a common engineering design choice for implementing control logic. Once the log characteristics are accurately identified, a POSITA would routinely store and retrieve corresponding optimal debarking settings from a database or lookup table, possibly with different settings for various intensity levels of characteristics (e.g., high moisture vs. low moisture logs). This is a predictable result of applying known control system design principles to the newly available, accurate upstream log data.

Conclusion:

Claims 1 and 11 of US12163947 would be rendered obvious by a combination of US 9,588,098 B2, WO2018/169712 A1, and US 10,099,400 B2, along with the general knowledge in the art. The motivation stems from the clear need to improve the accuracy and timeliness of log characterization for debarking optimization, the known superiority of deep learning for image analysis, and the desire to transition from reactive, downstream debarker control to proactive, upstream, real-time adjustment based on readily available, accurate log characteristics. The specific elements of the system and method, including contactless sensors, various log characteristics, and the use of lookup tables for parameters, represent routine applications and engineering choices given the primary combination.