Excavator load weighing system
The excavator load weighing system uses a camera and hydraulic pressure transducer to determine spatial bucket position and calculate weight, addressing the impracticality and cost of conventional systems, ensuring accurate and efficient load measurement.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- SIMPSON BRENT
- Filing Date
- 2025-12-23
- Publication Date
- 2026-07-09
Smart Images
Figure AU2025051490_09072026_PF_FP_ABST
Abstract
Description
Excavator Load Weighing SystemField of the Invention
[0001] This invention relates to excavator systems, and more particularly to systems and methods for determining the weight of materials loaded into an excavator bucket using spatial and hydraulic data.Background of the Invention
[0002] Excavators are widely used in industries such as construction, mining, and agriculture for tasks involving the movement and loading of materials. Accurate weight measurement of materials loaded into the excavator bucket is essential for efficient operations and safety. Overloaded trucks can lead to instability, structural damage, braking inefficiencies, and legal penalties. Similarly, underloading may result in inefficient operations and lost revenue, particularly in scenarios where payment is based on weight.
[0003] Conventional excavator load weighing systems typically rely on multiple transducers attached to various points on the boom and bucket to determine spatial positioning. These systems often require extensive calibration to account for bucket orientation and rely on hydraulic pressure measurements to calculate load weight. The complexity and high cost of such systems make them impractical for retrofitting existing excavators. Additionally, these systems can require significant installation time and may lack adaptability to various bucket configurations.
[0004] There remains a need for a cost-effective, adaptable, and accurate load weighing system that can be retrofitted to excavators with minimal modifications, ensuring precise weight measurements for safe and efficient operations..Summary of the Disclosure
[0005] The invention provides an excavator load weighing system comprising a camera, a hydraulic pressure transducer, a display, and a processor. The camera captures images of the excavator bucket, while the hydraulic pressure transducer measures hydraulic lift pressure. The processor determines the spatial position of the bucket relative to the camera using the captured images and calculates the weight ofmaterial loaded in the bucket based on the spatial position and the hydraulic lift pressure. The calculated weight is displayed on the display to provide real-time feedback to the operator.
[0006] This system is particularly suited for retrofitting to existing excavators, providing a cost-effective solution for accurate load measurement and enhancing operational safety and efficiency.
[0007] According to one aspect, there is provided an excavator load weighing system comprising a camera configured to capture images of an excavator bucket, a hydraulic pressure transducer configured to measure hydraulic lift pressure, a display, and a processor configured to determine a spatial position of the bucket relative to the camera using the captured images, to calculate a weight of material loaded within the bucket based on the spatial position and the hydraulic lift pressure, and to present the calculated weight on the display. Advantageously, this configuration enables accurate load measurement using a reduced sensor set, thereby facilitating cost-effective retrofit to existing excavators without requiring multiple positional transducers.
[0008] In some embodiments, the display may be configured to generate an alert when the calculated bucket weight exceeds a predefined threshold. This provides a technical advantage by enabling real-time operator feedback to reduce the risk of overload conditions and to improve operational safety.
[0009] In some embodiments, the processor may be configured to calculate a cumulative load weight by aggregating the calculated weights of successive bucket loads, and to generate an alert when the cumulative load weight exceeds a predefined threshold. This arrangement is particularly advantageous for vehicle loading operations, as it enables controlled loading to target weights while reducing reliance on external weighbridge systems.
[0010] In some embodiments, determination of the spatial position of the bucket relative to the camera may comprise identifying edges of the bucket within the captured images, calculating an apparent distance between the identified edges, and comparing the apparent distance to a known physical width of the bucket. Preferably,side edges of the bucket may be used for this determination, as the apparent spacing between side edges remains substantially invariant to changes in bucket pitch or tilt. This improves positional accuracy and reduces sensitivity to bucket orientation during lifting and dumping operations.
[0011] In some embodiments, the processor may be configured to perform image recognition to identify a type of bucket or attachment present within the camera field of view and to retrieve corresponding geometric reference data from a database. This enables automatic adaptation of the load calculation to different bucket types without manual recalibration, thereby improving usability and reducing setup time when attachments are changed.
[0012] In some embodiments, the system may be configured to calibrate or update a stored bucket width by capturing an image of the bucket positioned at a predefined distance from the camera and adjusting the stored width based on the apparent distance measured in the captured image. This calibration approach allows ongoing correction for dimensional tolerances, wear, or camera mounting variations, thereby improving long-term measurement accuracy.
[0013] In some embodiments, the processor may be configured to detect emptying of the bucket, to determine whether a calculated residual weight remains after emptying, and to automatically recalibrate the system when a non-zero residual weight is detected. Detection of emptying may be based on changes in apparent bucket geometry, such as variations in apparent height between front and rear edges, or on recognised motion patterns. This provides the technical benefit of mitigating cumulative drift and maintaining consistent zero reference during repetitive loading cycles.
[0014] In some embodiments, recalibration of the system may additionally or alternatively be manually initiated by an operator, or may be performed using an object of known weight placed in the bucket, with calibration parameters adjusted to align the calculated weight with the known weight. This allows flexible correction under varying site conditions and provides traceable calibration capability.
[0015] In some embodiments, the processor may be configured to calculate bucket load weight exclusively during detected lifting operations. Lifting detection may be based on hydraulic pressure thresholding, detection of drop-valve states, visionbased motion analysis, or combinations thereof. Restricting weight calculation to lifting states improves accuracy by avoiding pressure artefacts associated with lowering, dumping, or idle conditions.
[0016] In some embodiments, the processor may be configured to determine and track an operational cycle status for the bucket, including states such as loading, lifting, grading, dumping, returning, or idle, based on image-derived motion analysis and hydraulic data. The cycle status may be used to segment bucket events, to gate weight calculations, and to increment bucket load counts, thereby enabling structured analysis of work cycles and operational efficiency.
[0017] In some embodiments, the system may be configured to record bucket load counts, calculated weights, cumulative weights, and associated contextual data for reporting purposes, including association with vehicle loading sessions or general earthworks operations. This enables automated generation of load reports, productivity metrics, and audit records derived directly from sensed operational data, without reliance on manual logging.
[0018] Other aspects of the invention are also disclosed.Brief Description of the Drawings
[0019] Notwithstanding any other forms which may fall within the scope of the present invention, preferred embodiments of the disclosure will now be described, by way of example only, with reference to the accompanying drawings in which:
[0020] Figure 1 shows a block-level system diagram of an excavator load weighing system.
[0021] Figure 2 illustrates the system applied to a conventional excavator, showing the placement of the bucket at the end of the arm and the arrangement of the camera.
[0022] Figure 3 shows the field of view of the bucket from the perspective of the camera, with the bucket oriented directly facing the camera.
[0023] Figure 4 shows the field of view of the bucket from the perspective of the camera, with the bucket in a tilted orientation.
[0024] Figure 5 shows an exemplary view captured by the camera during operation of the excavator load weighing system, illustrating a bucket within the camera field of view with overlaid real-time operational parameters including bucket type, calculated load weight, cumulative total weight, bucket count, spatial distance measurement, and operational cycle status.Description of Embodiments
[0025] Figure 1 shows a block-level system diagram of an excavator load weighing system 100. The system 100 comprises a controller 104 that includes a processor configured to process digital data. The processor is in operable communication with a memory device and is configured to fetch digital data and computer program code instructions for decoding and executing the functionality described herein. These instructions may be logically divided into multiple program modules, each assigned specific subtasks.
[0026] The controller 104 includes an input / output (I / O) interface for connecting various peripherals and for sending and receiving data. In this regard, the system 100 comprises a camera 101 configured to capture a continuous stream of images of an excavator bucket 102 during operation.
[0027] Figure 2 illustrates the system 100 applied to a conventional excavator 106 equipped with the bucket 102 at the end of its arm. The arm typically includes a boom 108 and a stick 107, which are controlled by a hydraulic system comprising a stick hydraulic ram and a boom hydraulic ram 109. The boom hydraulic ram 109 is responsible for lifting the arm, and consequently, the bucket 102.
[0028] As shown in Figure 1, the controller 104 is operably connected to a hydraulic pressure sensor 105, which monitors the hydraulic ram 103 used for lifting the bucket 102. The hydraulic ram 103 typically corresponds to the boom hydraulic ram 109.
[0029] The camera 101 is mounted at a fixed reference point relative to the bucket 102, such as on the exterior or interior of the excavator's cabin 110. This placement ensures a consistent field of view of the bucket 102 during use. The system 100 isparticularly suitable for cost-effective retrofitting to existing excavators 106, providing weight measurement capabilities without significant modifications. Specifically, the controller 104 can be provided as a standalone device, mounted within the cabin 110, and powered by the excavator's electrical supply. The controller 104 connects to the installed camera 101 via a suitable data cable and interfaces with the pressure transducer 105 to obtain hydraulic pressure readings from the hydraulic ram 103. In some embodiments, the controller 104 also interfaces with a display.
[0030] In embodiments, the controller 104 may provide a network interface enabling a client device to access a user interface for data entry and reporting. For example, the controller 104 may host a lightweight web interface accessible via a browser application on a mobile phone, thereby allowing an operator to select a vehicle record, view the current bucket load count and calculated weights, and initiate report generation, while the controller 104 continues to perform the image processing and weight calculation locally.
[0031] During operation, the camera 101 captures a real-time stream of images of the excavator bucket 102, while the hydraulic pressure transducer 105 measures the hydraulic lift pressure of the hydraulic ram 103. The processor processes the captured images to determine the spatial position of the bucket 102 relative to the camera 101 , including the distance of the bucket 102 from the camera 101.
[0032] Figure 5 illustrates an exemplary view captured by the camera 101 during operation of the system, showing the excavator bucket within the camera field of view together with a set of real-time operational parameters generated by the processor 104. As shown in this figure, the processor 104 may be configured to maintain and update a plurality of operational parameters derived from combined image analysis and hydraulic pressure data. These parameters may be calculated continuously or at predefined sampling intervals and may be presented as overlays on the captured image stream or via a separate display interface. The tracked parameters may include an identified bucket or attachment type, a current calculated load weight, a cumulative total weight aggregated across multiple bucket cycles, a bucket load count, a measured spatial distance associated with the bucket position, and a currentoperational cycle status indicating a detected phase of operation such as lifting, dumping, loading, grading, returning, or idle.
[0033] Bucket or attachment identification may be performed by the processor 104 using the image recognition techniques previously described, with the resulting bucket type corresponding to an entry stored in the system database. Example bucket types may include mud buckets, general-purpose buckets, rock buckets, or other excavation or processing attachments. The identified bucket type may be used to select appropriate geometric reference data and calibration parameters and may additionally be stored as contextual metadata together with calculated weights and cycle records.
[0034] A current weight parameter may be calculated by the processor 104 to represent the instantaneous resolved weight of material contained within the bucket during a detected lifting operation. This current weight may be updated dynamically as the spatial position of the bucket or the measured hydraulic lift pressure changes and may correspond to the load associated with a single bucket cycle. Once a bucket load cycle is completed, the calculated weight may be stored and the current weight parameter reset in preparation for the next cycle.
[0035] The processor 104 may further maintain a total weight parameter representing a cumulative aggregate of the calculated weights from successive bucket loads. Each time a bucket load cycle is finalised, the processor may increment the total weight by the corresponding calculated bucket weight. The total weight may therefore increase progressively during a loading or earthworks operation and may be associated with a particular vehicle loading session or a general bulk material movement session.
[0036] A bucket count parameter may also be maintained to represent the number of discrete bucket load cycles detected during a session. The bucket count may be incremented upon detection of a completed cycle, for example following detection of bucket emptying and subsequent return or repositioning of the bucket. The bucket count may be stored together with associated weight values and time stamps to support subsequent reporting and operational analysis.
[0037] Spatial measurements derived from image analysis may be used to calculate and display a distance-related parameter, such as a cutting distance or a distance between the bucket and a reference point. This distance parameter may be derived from the apparent width of the bucket within the camera field of view and converted to a real-world measurement using the known bucket geometry and camera parameters. The distance parameter may be used internally for load calculation and may also be presented as an operational metric during lifting, grading, or dumping activities.
[0038] The processor 104 may determine a cycle status parameter representing the current operational state of the bucket within a predefined cycle model. The cycle status may be inferred from detected motion patterns of the bucket within the image stream, spatial position changes over time, and variations in hydraulic pressure. Example cycle states may include loading, lifting, grading, dumping, returning, and idle, with transitions between states determined by threshold conditions or recognised movement sequences.
[0039] Cycle status information may be used by the processor 104 to control when weight calculations are performed and to segment recorded data into discrete bucket cycles. For example, weight calculation may be enabled during lifting states and inhibited during non-lifting states, while detection of dumping states may trigger storage of the calculated bucket weight and incrementation of the bucket count. The cycle status may also be stored as part of each bucket event record, enabling later analysis of cycle efficiency, operational patterns, and productivity metrics.
[0040] The weight of the material loaded within the bucket 102 is calculated by combining this spatial data with the hydraulic lift pressure measured by the transducer 105. For example, if the bucket 102 is positioned 3 metres away from the camera 101 and the hydraulic pressure reading indicates 200 bar, the processor calculates the force exerted by the hydraulic ram 103 based on the measured pressure. This force may be adjusted using the calculated moment arm, derived from the distance of the bucket 102 from the camera 101 and the known pivot point geometry of the excavator arm. For example, the processor may calculate the torque applied by the hydraulicram 103 and resolves it into a vertical lifting force. This force is then correlated with the load weight, factoring in the bucket’s spatial position and the known dimensions of the arm and bucket. The calculated weight may be displayed on the display to provide real-time feedback to the operator.
[0041] The controller 104 may be further configured to monitor and display alerts to the operator based on predefined weight thresholds. For example, the processor may compare the calculated weight of material in the bucket 102 to a predefined threshold stored in the system’s memory. When the calculated weight exceeds this predefined threshold, the processor may send a signal to the display to present an alert to the operator. The alert may include visual indicators, such as flashing text, a warning icon, or colour-coded displays, and may also include auditory signals to ensure the operator’s immediate attention.
[0042] Additionally, the processor may be configured to calculate a cumulative load weight by summing the weights of individual bucket loads during a loading operation. As each load is calculated, the processor may update the cumulative weight in realtime, allowing the operator to monitor progress towards a target weight, such as the maximum permissible load of a truck.
[0043] The system 100 may also be configured to compare the cumulative load weight against a predefined cumulative threshold, which may correspond to the safe or legal weight limit of a truck or other receiving vehicle. When the cumulative load weight approaches or exceeds this threshold, the processor may send a signal to the display to present an alert. This cumulative weight alert may include visual and auditory warnings, such as displaying a "load limit reached" message, changing the colour of the weight indicator to red, or playing an audible warning tone.
[0044] In embodiments, the controller 104 may maintain a vehicle database stored in the memory device, the vehicle database comprising one or more vehicle records. Each vehicle record may include a vehicle identifier such as a vehicle registration number and may optionally include associated metadata such as a driver name and a driver contact number. The vehicle database may be populated by an operator via a user interface presented on the display, or via a remote client device incommunication with the controller 104, for example a mobile phone browser session that accesses a simplified operator webpage served by the controller 104.
[0045] In embodiments, the controller 104 may be configured to manage discrete loading sessions, each loading session being associated with a selected vehicle record. During a loading session, the processor may increment a bucket load count and store, for each bucket event, a calculated bucket load weight together with a time stamp, the identified vehicle record, and optionally an identified bucket type. The controller 104 may thus store per-vehicle aggregates including a number of bucket loads, a total tonnage, and optionally statistical measures such as average bucket load weight and weight variance, thereby enabling post-operation reconciliation and reporting.
[0046] In embodiments, the controller 104 may be configured to generate an end-of-shift or end-of-day report using the stored loading session data. The report may be generated in a machine-readable format such as a comma-separated values file or JavaScript Object Notation, and may additionally be rendered as a human-readable summary on the display. By way of example, the report may include, for each vehicle record loaded during the reporting period, the vehicle identifier, the number of bucket loads recorded, and the total tonnage calculated from the stored bucket load weights, and may optionally include the time window of the loading session and any threshold alerts triggered during that session.
[0047] In embodiments, the controller 104 may be configured to support optional association of a vehicle record with a remote device identifier that is observed during loading operations, thereby reducing operator interaction. For example, the controller 104 may detect one or more wireless device identifiers present proximate the excavator cabin 110 during a loading session, such as a Bluetooth identifier or a WiFi identifier broadcast by a driver device, and may store an association between the observed identifier and a corresponding vehicle record. In use, when the controller 104 observes the same identifier during a subsequent loading session, the processor may automatically preselect the corresponding vehicle record for operator confirmation prior to commencing storage of per-bucket records for that session.
[0048] In embodiments, the system 100 may optionally support vehicle identification using additional sensing hardware, such as a second camera positioned to capture images of a receiving vehicle, wherein the processor performs number plate recognition on images captured by the second camera to determine a vehicle identifier. Where provided, a recognised vehicle identifier may be used to retrieve or create a vehicle record in the vehicle database and to associate the loading session data with that vehicle record. In such embodiments, the system may store both the vehicle identifier and a confidence score for the recognition result, and may prompt the operator for confirmation if the confidence score is below a predefined threshold.
[0049] In embodiments, the controller 104 may be configured to operate in a general earthworks tracking mode in which bucket events are recorded and aggregated without association to a vehicle record. In this mode, the processor may store, for each bucket event, the calculated bucket load weight, a time stamp, and optionally the spatial position of the bucket relative to the camera 101 at the time of the weight calculation, and may compute daily totals including the number of bucket events recorded and a total tonnage moved.
[0050] In embodiments, the controller 104 may compute performance metrics derived from the stored bucket event data to enable operational auditing. By way of example, the processor may compute a bucket events-per-hour rate, a total-tonnage-per-hour rate, and time-segmented aggregates (for example, per hour or per predefined work interval) based on the time stamps stored for bucket events. The controller 104 may store these metrics as part of the end-of-day report and may optionally present them on the display as trend indicators over a reporting period. In some embodiments, the controller 104 may export the daily totals for comparison against external volumetric or surveyed estimates, for example by allowing a supervisor system to import the generated report.
[0051] Figures 3 and 4 show a field of view 111 of the bucket 102 from the perspective of the camera 101. In embodiments, the processor may be configured to determine the position of the bucket 102 relative to the camera 101 by employing image processing techniques to identify the edges 113 of the bucket 102 within the capturedimages. For edge identification, the processor may use gradient-based methods, such as the Sobel operator or Canny edge detection, to detect areas in the image where pixel intensity changes sharply, corresponding to the edges of the bucket. These techniques may include pre-processing steps, such as Gaussian smoothing, to reduce noise in the image and improve edge detection accuracy.
[0052] Once the edges 113 of the bucket 102 are identified, the processor may further refine the detected edges using contour analysis techniques, such as the Hough Transform, to extract linear or curved segments that correspond to the actual edges of the bucket. The processor may then use filtering techniques to remove false positives, such as edges from the surrounding environment or excavator components, by focusing on geometric features known to correspond to the bucket's structure.
[0053] To calculate the apparent distance between the identified edges, the processor may apply pixel-to-distance mapping based on the resolution and focal length of the camera. By measuring the number of pixels between the identified edges and converting this pixel distance into a real-world measurement, the processor may determine the apparent width 111 of the bucket 102 as it appears in the image. This calculation may also involve perspective correction techniques, such as homography, to account for distortions caused by the camera's angle relative to the bucket.
[0054] The processor may compare the calculated apparent width of the bucket 102 to its known physical width, which may be preloaded into the system's memory. Using this comparison, the processor may estimate the bucket's distance from the camera 102. This positional information may be further refined through additional algorithms, such as least squares optimisation, to minimise errors and improve accuracy.
[0055] In a preferred embodiment, the processor 104 may be configured to use the side edges 113A of the bucket 102, as opposed to the front and rear (or long) edges 113B, to determine the distance of the bucket 102 from the camera 101. This selection is based on the observation that the angle of the bucket 102 relative to the stick 107 can affect the apparent height 111 B of the bucket 102, as measured between the long edges 113B, independently of the bucket's distance from the camera 101. Suchangular changes can distort measurements based on the long edges, introducing variability that is unrelated to the bucket's spatial position.
[0056] In contrast, the apparent width 111A between the side edges 113A remains invariant to the angle of the bucket 102 relative to the stick 107 or the camera's field of view 111. This invariance ensures that measurements derived from the side edges 113A provide a consistent and reliable basis for determining the bucket's distance from the camera 101. For example, as shown in Figure 3, the bucket 102 is oriented directly facing the camera 101, and the apparent width 111 A is measured between the side edges 113A. Figure 4 illustrates the bucket 102 in a tilted orientation, with its opening facing upward relative to the camera's field of view 111. Despite this angular change, the apparent width 111A between the side edges 113A remains unaffected.
[0057] This approach leverages the geometric consistency of the side edges 113A to mitigate potential errors caused by changes in the bucket's pitch or tilt angle. By relying on the invariant apparent width 111A, the processor 104 can accurately calculate the bucket's distance from the camera 101 regardless of its angular orientation. The calculation may involve pixel-to-distance mapping, as described earlier, which uses the measured apparent width in pixels and converts it into a real-world distance based on the known physical width of the bucket 102 and the camera's intrinsic parameters.
[0058] In embodiments, the processor 104 may be configured to perform image recognition to visually identify the type of bucket 102 in the captured images. This identification process may involve using advanced image recognition techniques or trained neural network models. The system may store a database of bucket types, each associated with specific physical characteristics, such as the known width of the bucket, for use in subsequent calculations.
[0059] The processor 104 may use conventional image recognition techniques, such as feature-based methods, including Scale-Invariant Feature Transform (SIFT) or Speeded-Up Robust Features (SURF), to extract unique features from the bucket 102. These features may then be matched against a pre-stored set of templates inthe database to identify the bucket type. Alternatively, histogram-based methods, such as Histogram of Oriented Gradients (HOG), may be employed to detect bucketspecific shapes and patterns.
[0060] In alternative embodiments, the processor 104 may employ a convolutional neural network (CNN) trained to recognise different bucket types. The CNN may be trained on a dataset comprising labelled images of various buckets captured from multiple angles, under varying lighting conditions, and in different operational environments. During operation, the CNN processes the captured images to classify the bucket 102 based on its visual features. The classification output may correspond to a specific bucket type stored in the database.
[0061] Once the bucket type is identified, the processor 104 may retrieve the known width of the identified bucket from the database. This known width is then used as a reference for subsequent calculations, such as determining the bucket’s position relative to the camera 101 or calculating the weight of the material loaded within the bucket.
[0062] In some embodiments, the processor 104 may also incorporate confidence scoring into the identification process. If the confidence score of the bucket classification is below a predefined threshold, the system may prompt the operator to confirm or manually select the bucket type from a list displayed on the user interface. This feature ensures robust operation even in scenarios where visual obstructions or environmental conditions could impact recognition accuracy.
[0063] In embodiments, the database stored in the memory device may further store attachment records corresponding to different excavator attachments that may be presented within the camera 101 field of view, including buckets and non-bucket attachments. Each attachment record may store one or more charge attributes associated with use of that attachment, such as a base charge category, an incremental surcharge flag, or a billing code, and may further store one or more technical attributes used by the processor for measurement calculations, such as a reference width or a geometric correction factor.
[0064] The processor may be configured to determine a current attachment state of the excavator by classifying captured images to identify which attachment is attached at a given time, and to detect transitions between attachment states. A transition may be detected, for example, when the classification output changes from a first attachment record to a second attachment record with a confidence score exceeding a predefined threshold for a predefined number of consecutive frames, thereby reducing false transitions due to transient occlusion or lighting variations. The controller 104 may store attachment state intervals as time-stamped records, thereby enabling allocation of measured work outputs (such as total tonnage and bucket event counts) to corresponding attachments.
[0065] The controller 104 may be configured to expose a time-sheet style summary derived from the attachment state intervals. For example, the controller 104 may calculate a total duration of use for each identified attachment during a reporting period and may associate that duration with the applicable charge attributes stored in the corresponding attachment record. The resulting summary may be exported together with the bucket event and loading session reports, thereby enabling downstream invoicing systems to apply different rates or surcharges depending on the attachment used during the recorded work intervals, without requiring manual reconstruction of attachment usage.
[0066] In embodiments, the processor 104 may be configured to perform a calibration routine to determine and update the stored width of the bucket 102. This calibration process may be initiated by capturing an image of the bucket 102 positioned at a predefined, known distance from the camera 101. This predefined distance may be established by placing the bucket 102 in a reference position, such as lowering the bucket to a ground-level mark or aligning it with a visible calibration target within the excavator's operational range.
[0067] During the calibration routine, the camera 101 captures an image of the bucket 102 in this predefined position, ensuring that the bucket is fully within the field of view 111. The processor 104 processes the captured image using edge detection techniques, such as the Sobel operator, Canny edge detection, or contour analysis,to identify the bucket’s side edges 113A. Once the side edges 113A are identified, the processor 104 calculates the apparent distance between the edges in terms of pixel width.
[0068] Using pixel-to-distance mapping, the processor 104 converts this apparent pixel width into a real-world measurement of the bucket’s width at the predefined distance. This conversion may leverage camera parameters, such as focal length and sensor resolution, along with the known distance of the bucket 102 from the camera 101. Perspective correction techniques, such as homography, may also be applied to account for any angular distortion caused by the relative alignment of the camera 101 and the bucket 102.
[0069] The processor 104 then compares the calculated width to the previously stored width value for the bucket 102 in the system’s memory. If discrepancies are identified, the processor 104 updates the stored width value with the newly calculated width, ensuring that subsequent measurements are based on accurate dimensional data. This updated width value may also be associated with the identified bucket type if the system includes image recognition capabilities for bucket classification.
[0070] In some embodiments, the system 100 may guide the operator through the calibration process via the display. For instance, the display may provide instructions to position the bucket at the predefined distance and confirm when the calibration image is successfully captured. Additionally, the system may verify the consistency of the calibration by analysing multiple images captured during slight movements of the bucket, refining the stored width value through averaging or statistical filtering to minimise errors.
[0071] In embodiments, the processor 104 may be configured to detect the emptying of the bucket 102 and use this detection to initiate a recalibration process if the calculated weight of the bucket remains nonzero after emptying. The detection of bucket emptying may rely on variations in the apparent height 111 B between the front and rear edges 113B of the bucket 102, as captured in the field of view 111 of the camera 101. Specifically, the processor 104 may monitor the apparent height of the bucket over time and recognise patterns indicative of material being emptied, suchas a rapid change in the height profile of the bucket 102, which corresponds to the release of its contents.
[0072] Alternatively, the processor may analyse hydraulic control signals to detect emptying of the bucket 102, such as retraction of the stick 107 hydraulic ram.
[0073] Upon detecting that the bucket has been emptied, the processor 104 may compare the calculated weight of the bucket 102 to a predefined zero-weight threshold stored in the system’s memory. If the calculated weight is nonzero beyond an acceptable margin of error, the processor 104 may automatically initiate a recalibration process. During recalibration, the system may reset its internal weight calculations to align with the known empty state of the bucket, thereby ensuring accurate subsequent measurements.
[0074] The processor 104 may also be configured to automatically initiate recalibration following specific bucket control signals. For example, the system may detect the sequence of bucket operations where the bucket is first emptied and then swung, a pattern indicative of the end of one loading cycle and preparation for the next. The processor 104 may recognise this sequence by interfacing with hydraulic control signals or by analysing image data for movement patterns consistent with bucket swinging. Once this sequence is detected, the processor 104 may confirm the bucket’s empty state and recalibrate the system before the next load is scooped.
[0075] In addition to automatic recalibration, the processor 104 may be manually operable to initiate recalibration. For instance, if the operator observes a discrepancy in the displayed weight after the bucket 102 has been emptied, the operator may press a calibration button on the user interface. This action signals the processor 104 to reset the weight calculation, ensuring accuracy without relying solely on automatic detection.
[0076] The combination of automatic and manual recalibration mechanisms enhances the system’s reliability and flexibility. Automatic recalibration ensures seamless operation by resetting the system after each load is dropped, minimising the risk of cumulative errors. Manual recalibration provides the operator with control to addressdiscrepancies in real time, especially in scenarios where environmental factors or operational irregularities may affect system performance.
[0077] In embodiments, the processor 104 may be configured to recalibrate the calculated weight by utilising an object of known weight placed in the bucket 102. This process begins with the operator selecting a calibration mode via the system’s user interface. Once the mode is activated, the operator places an object with a predefined weight, such as a standard concrete pipe commonly found on construction sites, into the bucket 102. The known weight of the object may be manually entered by the operator into the system through the interface or selected from a pre-stored database of calibration objects.
[0078] During calibration, the processor 104 calculates the weight of the object based on the bucket's position and the hydraulic lift pressure data, using the system’s standard weight computation algorithms. The calculated weight is then compared to the known weight of the calibration object entered or retrieved by the system. If a discrepancy is detected between the calculated weight and the known weight, the processor 104 adjusts the system’s calibration parameters to minimise the discrepancy.
[0079] In embodiments, the processor 104 may be configured to calculate the weight of material loaded within the bucket 102 exclusively during lifting operations. This ensures that weight measurements are taken under consistent conditions, avoiding inaccuracies caused by bucket lowering or other non-lifting movements that may affect hydraulic pressure readings.
[0080] To detect lifting of the bucket 102, the processor 104 may monitor the state of a drop valve associated with the hydraulic system. A drop valve is typically deactivated when the hydraulic ram 103 is pressurised to lift the bucket and activated when the bucket is stationary or positioned to prevent unintended movement. The processor 104 may interface with hydraulic control signals to detect deactivation of the drop valve, thereby identifying the initiation of a lifting operation. This deactivation serves as a trigger for the processor 104 to calculate the weight of the bucket's contents.
[0081] Additionally, the processor 104 may use hydraulic pressure thresholding to detect lifting of the bucket 102. Specifically, the processor 104 may analyse pressure readings from the hydraulic pressure transducer 105 and compare these readings to a predefined threshold stored in the system's memory. If the measured pressure exceeds the threshold, this indicates that the hydraulic ram 103 is actively lifting the bucket. The processor 104 may combine this threshold detection with the drop valve deactivation signal for improved accuracy and reliability in identifying lifting operations.
[0082] In embodiments where both methods are employed, the processor 104 may detect lifting of the bucket 102 by confirming both the deactivation of the drop valve and the crossing of the hydraulic pressure threshold. This dual-condition detection reduces the likelihood of false positives caused by transient pressure changes or control signal anomalies. By requiring both conditions to be satisfied, the system ensures that weight calculations are performed only when the bucket is definitively in a lifting state.
[0083] Once lifting is detected, the processor 104 calculates the weight of the bucket's contents based on the hydraulic lift pressure and the spatial position of the bucket relative to the camera 101. The calculated weight is then displayed on the user interface to inform the operator.
[0084] In embodiments, instead of relying solely on identifying the edges of the bucket 102, the processor 104 may employ image processing techniques to ascertain the corners of the bucket. By identifying corner points, the processor 104 can determine both the apparent width and height of the bucket relative to the camera 101. Corner detection may utilise methods such as the Harris corner detector, Shi-Tomasi corner detection, or deep learning-based models trained on bucket corner features. The detected corner points can then be used to calculate the bucket’s spatial orientation and size, providing an alternative approach to measuring the apparent width and height.
[0085] The system may also accommodate buckets 102 that tilt sideways, such as tilt buckets and tilt hitches. A tilt bucket typically includes hydraulic rams that allow thebucket 102 to tilt at varying angles, enabling more versatile excavation tasks. A tilt hitch, on the other hand, is a permanent quick hitch that allows attachments, including buckets 102, to be tilted at adjustable angles. Even when such attachments alter their angles, the system’s vision analysis capabilities can determine the apparent width and height of the bucket by correcting for angular distortion. Techniques such as homography transformations, 3D pose estimation, or a combination of depth mapping and perspective correction can be used to accurately measure the bucket’s dimensions regardless of its orientation.
[0086] While hydraulic pressure readings can be used to detect the raising or lowering of the bucket (e.g., pressure exceeding a predefined threshold may indicate lifting), in preferred embodiments, the detection of raising or lowering is achieved using vision analysis alone. This may involve monitoring the movement of the bucket within the camera’s field of view over a sequence of images. For instance, optical flow algorithms, such as the Lucas-Kanade method or Farneback dense optical flow, can track the motion of bucket features between successive frames to detect lifting or lowering movements. The vertical displacement of the bucket’s edges or corners relative to the camera can be calculated to ascertain its motion direction, eliminating the need to interface with hydraulic systems.
[0087] Regarding drop valves, these are safety-critical components within the hydraulic system and are preferably not interfaced directly by the processor 104. Instead, the system 100 preferably relies on non-invasive methods, such as visionbased or indirect hydraulic signal analysis, to detect bucket operations. This approach ensures that the load weighing system does not interfere with or compromise the safety functions of the drop valves, maintaining the integrity of the excavator’s hydraulic safety mechanisms. Vision-based detection, for example, can identify lifting, lowering, or tilting operations without requiring direct interaction with hydraulic control signals.
[0088] An exemplary method of use for the excavator load weighing system 100 may begin with the operator calibrating the system prior to operation. Calibration may involve positioning the bucket 102 at a predefined reference point, such as loweringit to a ground-level marker or aligning it with a visible calibration target. The processor 104 may then capture an image of the bucket 102 using the camera 101, identify its side edges 113A through edge detection techniques, and calculate the apparent width of the bucket in the image. The processor 104 may compare this calculated width to the known physical width stored in the system’s memory and update the stored width value if discrepancies are detected.
[0089] During material loading, the operator may fill the bucket 102 with material using the excavator's controls. The camera 101 may continuously capture images of the bucket 102, and the hydraulic pressure transducer 105 may measure the hydraulic lift pressure of the hydraulic ram 103. The processor 104 may process the captured images to determine the bucket's spatial position relative to the camera 101 and use this position along with the hydraulic pressure data to calculate the weight of the material loaded within the bucket. The calculated weight may be displayed in realtime on the display to inform the operator.
[0090] If the calculated weight exceeds a predefined threshold stored in the system’s memory, the processor 104 may send a signal to the display to present an alert. This alert may take the form of visual indicators, such as a warning icon or flashing text, or auditory signals to draw the operator’s attention. The operator may continue loading material into a truck, and the processor 104 may sum the weights of individual bucket loads to calculate a cumulative load weight. If the cumulative load weight approaches or exceeds a predefined threshold, the processor 104 may display an additional alert to indicate that the truck is fully loaded.
[0091] The processor 104 may also automatically detect when the bucket 102 is emptied by monitoring variations in the apparent height 111 B between the front and rear edges 113B of the bucket 102, as captured in the camera’s field of view 111. If the processor 104 determines that the calculated weight of the bucket 102 remains nonzero after emptying, it may automatically recalibrate the system by resetting internal weight calculations to align with the known empty state of the bucket.
[0092] The operator may also initiate a manual recalibration if a discrepancy is observed. For example, if the operator places an object of known weight into thebucket 102, they may enter the known weight into the system via the user interface. The processor 104 may then compare the calculated weight of the object to the entered weight and adjust the system’s calibration parameters to minimise any discrepancy. Once recalibration is complete, the system is ready for continued operation.
[0093] Throughout the loading process, the processor 104 may ensure that weight calculations are performed exclusively during lifting operations. This may involve detecting the deactivation of a drop valve or monitoring hydraulic pressure readings to determine when the bucket 102 is being lifted. These capabilities ensure that the system delivers accurate and consistent weight measurements during operation, supporting safe and efficient loading practices.
[0094] The foregoing description, for purposes of explanation, used specific nomenclature to provide a thorough understanding of the invention. However, it will be apparent to one skilled in the art that specific details are not required in order to practise the invention. Thus, the foregoing descriptions of specific embodiments of the invention are presented for purposes of illustration and description. They are not intended to be exhaustive or to limit the invention to the precise forms disclosed as obviously many modifications and variations are possible in view of the above teachings. The embodiments were chosen and described in order to best explain the principles of the invention and its practical applications, thereby enabling others skilled in the art to best utilize the invention and various embodiments with various modifications as are suited to the particular use contemplated. It is intended that the following claims and their equivalents define the scope of the invention.
Claims
Claims1. An excavator load weighing system comprising:a camera configured to capture an image of an excavator bucket;a hydraulic pressure transducer configured to measure hydraulic lift pressure; a display; anda processor configured to:(i) process the captured image to determine a spatial position of the bucket relative to the camera,(ii) use the spatial position and hydraulic lift pressure data to calculate the weight of material loaded within the bucket, and(iii) display the weight on the display.
2. The system of claim 1, wherein the display is configured to display an alert when the weight exceeds a predefined threshold.
3. The system of claim 1, wherein the processor is further configured to calculate a cumulative load weight.
4. The system of claim 3, wherein the display is configured to display an alert when the cumulative load weight exceeds a predefined threshold.
5. The system of claim 1, wherein the processor is configured to determine the position of the bucket relative to the camera by:identifying edges of the bucket in the captured image,calculating the apparent distance between the edges, andcomparing the apparent distance to the known width of the bucket.
6. The system of claim 5, wherein the processor is configured to use the side edges of the bucket to determine the position of the bucket relative to the camera.
7. The system of claim 5, wherein the processor is configured to perform image recognition to identify the type of bucket in the captured images and to retrieve the known width of the identified bucket from a database.
8. The system of claim 5, wherein the processor is configured to calibrate a width of the bucket by capturing an image of the bucket positioned at a predefined distance from the camera and updating the stored width value based on the apparent distance between the edges of the bucket in the captured images.
9. The system of claim 1, wherein the processor is configured to detect the emptying of the bucket, determine if the calculated weight is nonzero after emptying, and automatically recalibrate the system if the calculated weight is nonzero.
10. The system of claim 9, wherein the processor is configured to automatically recalibrate the system upon determining that the calculated weight of the bucket is nonzero after detecting the emptying of the bucket.
11. The system of claim 9, wherein the processor is configured for detecting emptying of the bucket according to variations in the apparent height between the front and rear edges of the bucket.
12. The system of claim 1, wherein the processor is manually operable to initiate recalibration.
13. The system of claim 1, wherein the processor is configured to automatically initiate recalibration responsive to bucket control signals by detecting the emptying of the bucket and subsequent swinging of the bucket.
14. The system of claim 1, wherein the processor is configured to recalibrate the calculated weight by receiving an input corresponding to a known weight of an objectplaced in the bucket, comparing the calculated weight to the known weight, and adjusting calibration parameters to align the calculated weight with the known weight.
15. The system of claim 1, wherein the processor is configured to calculate the weight exclusively when detecting the lifting of the bucket.
16. The system of claim 15, wherein the processor is configured to detect deactivation of a drop valve to determine when the bucket is being lifted.
17. The system of claim 15, wherein the processor is configured to threshold hydraulic pressure readings to detect when the bucket is being lifted.
18. The system of claim 17, wherein the processor is configured to detect deactivation of a drop valve and to threshold hydraulic pressure readings to determine when the bucket is being lifted.
19. A method of weighing a load in an excavator bucket, the method comprising: capturing an image of the bucket using a camera;measuring hydraulic lift pressure of the bucket using a hydraulic pressure transducer;determining a spatial position of the bucket relative to the camera by processing the captured image;calculating a weight of material loaded in the bucket based on the spatial position and the hydraulic lift pressure; anddisplaying the weight on a display.
20. The method of claim 19, further comprising displaying an alert on the display when the weight of material in the bucket exceeds a predefined threshold.
21. The method of claim 20, further comprising:calculating a cumulative load weight by summing individual bucket load weights during a loading operation; anddisplaying an alert on the display when the cumulative load weight exceeds a predefined threshold.
22. The method of claim 19, further comprising:identifying edges of the bucket in the captured image;calculating an apparent distance between the identified edges;comparing the apparent distance to a known width of the bucket to determine its position relative to the camera; andusing the determined position to calculate the weight of the material in the bucket.
23. The method of claim 22, further comprising using the side edges of the bucket to determine its position, wherein the side edges provide a consistent reference independent of the bucket's angle.
24. The method of claim 19, further comprising:detecting the emptying of the bucket by monitoring variations in the apparent height between the front and rear edges of the bucket;determining if the calculated weight of the bucket is nonzero after emptying; and automatically recalibrating the system if the calculated weight is nonzero.
25. The method of claim 19, further comprising:receiving an input corresponding to a known weight of an object placed in the bucket;comparing the calculated weight of the object to the known weight; and adjusting calibration parameters to align the calculated weight with the known weight.
26. The method of claim 19, further comprising calculating the weight of the material in the bucket exclusively during lifting of the bucket by detecting:deactivation of a drop valve;hydraulic pressure readings exceeding a predefined threshold; orboth deactivation of the drop valve and hydraulic pressure readings exceeding the predefined threshold.
27. The method of claim 19, further comprising:automatically initiating recalibration by detecting the sequence of the bucket being emptied and subsequently swung; andresetting weight calculations to align with the empty state of the bucket.