State-Based Payload Capture for Work Machine Productivity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional onboard payload scales in work machines provide inaccurate and messy bucket weight measurements due to pile impacts and non-productive work activities, making it difficult to accurately represent productivity.
Innovation Solution
A state-based payload capturing technique using machine learning and onboard sensors to detect event-based transitions between work states, selectively capturing payload data only during defined periods of productive work, thereby distinguishing between productive and non-productive activities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional onboard payload scales continuously capture bucket weight data, then all payload measurements are recorded, but the data becomes messy and inaccurate due to pile impacts and non-productive work
Solution Approach 1:
The work cycle is segmented into distinct work states (digging, transporting, dumping, idle) using machine learning classification of sensor data. Payload measurements are then captured selectively only during productive work states, separating useful productivity data from noise generated during non-productive activities like pile impacts or idle movements.
Solution Approach 2:
A machine learning-based work state estimation system acts as an intermediary between the raw payload scale data and the final productivity measurement. This intermediary classifies the current work state based on multiple sensor inputs and selectively enables payload capture only when productive work is detected, filtering out inaccurate measurements without losing meaningful productivity information.
2Productivity
If payload data is captured during all work activities, then complete data coverage is achieved, but productivity representation becomes inaccurate due to inclusion of non-productive work
Solution Approach 1:
The payload capture system dynamically adjusts its operation based on the detected work state. Instead of continuous static capture, the system transitions between capture and non-capture modes according to the real-time classification of work activities, ensuring that only payload data from productive operations is recorded, thereby improving both productivity measurement and data reliability.
Solution Approach 2:
The system uses feedback from multiple sensors (accelerometers, GPS, hydraulic pressure sensors) to continuously monitor and classify the current work state. This feedback loop enables the system to distinguish between productive and non-productive activities in real-time, adjusting payload capture accordingly to ensure high reliability of the recorded productivity data.
3Measurement precision
If machine learning-based work state detection is implemented, then selective payload capture during productive work is enabled, but system complexity increases
Solution Approach 1:
The machine learning work state estimation system serves multiple functions simultaneously: it classifies work states for selective payload capture, provides operational context for productivity analysis, and enables various reporting capabilities. This multi-functionality justifies the added complexity by delivering comprehensive insights beyond simple payload measurement.
Solution Approach 2:
The system uses the work machine's existing sensors and operational data to automatically classify work states and control payload capture without requiring external intervention or complex additional hardware. The machine essentially serves itself by leveraging its own operational information to improve measurement accuracy.
Data Source
AI summary
A computer-implemented system and method are provided for state-based payload capture by a work machine comprising ground engaging units supporting a frame, and a work implement moveable with respect to the frame for loading and unloading payloads. At least a first set of sensors associated with the work machine are used to detect event-based transitions between work states in a defined work cycle having a sequence of work states therein, e.g., from digging states to loaded states. An onboard payload measuring unit is used to selectively capture payload data corresponding to a current work cycle in association with the detected transition. The captured payload data is categorized and stored, independently for the current work cycle with respect to associated locations within a work site and/or with respect to time, and in aggregate with other captured payload data for each of a plurality of work cycles for the work machine.