Work Machine Control Using Real-Time Predictive Yield Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for controlling work machines, such as combine harvesters, rely on a priori data like aerial imagery to generate predictive yield maps, which do not account for actual ground truth data and are only improved post-harvest, leading to inefficiencies in real-time operation.
Innovation Solution
A control system that generates a predictive model using both georeferenced vegetative index data and in situ field data collected during the operation, dynamically evaluating and updating the model for improved accuracy and control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If a priori data (aerial imagery) is used to generate predictive yield maps, then the control system can operate with pre-existing information, but the model accuracy does not reflect actual ground truth conditions
Solution Approach 1:
The system performs preliminary actions by collecting and processing field data during harvest operations to generate predictive yield maps in advance. Sensors on the harvester collect crop data, moisture content, and yield information while moving through the field, allowing the model to be trained with actual ground truth data before final harvest completion, thus improving both timing and accuracy
Solution Approach 2:
The system implements feedback by continuously comparing predicted yield values with actual measured yield data from sensors during harvest. This feedback loop allows the predictive model to be dynamically updated and refined in real-time, ensuring the model accuracy reflects actual ground truth conditions while maintaining operational efficiency
2Measurement precision
If the predictive model is dynamically updated during operation, then the model accuracy improves, but the system complexity increases
Solution Approach 1:
The control system achieves multi-functionality by integrating multiple roles into a single system: data collection from sensors, real-time model training, model validation, and harvest control. This universal approach improves predictive accuracy without proportionally increasing system complexity, as one integrated system performs what would otherwise require multiple separate subsystems
Solution Approach 2:
The predictive model performs self-service by automatically training and updating itself using data collected during harvest operations. The system autonomously processes field data, adjusts model parameters, and validates predictions without requiring external intervention, thereby improving accuracy while minimizing the operational burden and complexity for users
3Measurement precision
If field data is collected during operation, then real-time model accuracy improves, but the data collection and processing time increases
Solution Approach 1:
The system maintains continuity of useful action by collecting, processing, and updating model data continuously during harvest operations without interrupting the harvesting process. Sensors continuously capture field data, the model continuously learns from this data, and predictions are continuously updated, ensuring real-time accuracy while minimizing time loss through seamless integration with ongoing operations
Solution Approach 2:
The system performs preliminary data processing and model updates in advance during the harvest operation itself, rather than waiting until after harvest is complete. By training the model incrementally as data becomes available during operations, the system prepares accurate predictions ahead of time, reducing the need for post-harvest processing and enabling real-time decision-making
Data Source
AI summary
A priori georeferenced vegetative index data is obtained for a worksite, along with field data that is collected by a sensor on a work machine that is performing an operation at the worksite. A predictive model is generated, while the machine is performing the operation, based on the georeferenced vegetative index data and the field data. A model quality metric is generated for the predictive model and is used to determine whether the predictive model is a qualified predicative model. If so, a control system controls a subsystem of the work machine, using the qualified predictive model, and a position of the work machine, to perform the operation.