Excavator Inertial Sensor Calibration via Multi-Point Posture Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional construction machines face challenges in accurately computing the posture of work implements like booms, arms, and buckets due to inherent errors in sensors, which reduce precision when working on surfaces different from the reference plane used during calibration, especially in environments without external measuring devices.
Innovation Solution
A construction machine equipped with a multipoint front work implement, posture information sensors, and a front posture computing device that sets a reference position, computes calibration parameters, and determines the relative position of the work tool, allowing for precise posture computation without external devices by capturing multiple postures of the driven members.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional single-point calibration is used to simplify the calibration process, then the calibration can be completed quickly and easily, but the posture computation precision deteriorates when working on surfaces different from the reference plane
Solution Approach 1:
The calibration process is segmented into multiple calibration points along the reference plane. Instead of using a single calibration point, the system collects posture data at multiple discrete points (e.g., multiple positions of the work tool along the longitudinal direction). This segmentation allows the system to capture sensor errors across different positions and compute correction values that are valid for various working surfaces, thereby maintaining high precision without complicating the overall calibration procedure.
Solution Approach 2:
The calibration approach transitions from single-point (zero-dimensional) to multi-point (one-dimensional along the reference plane). By adding the dimension of multiple calibration positions, the system gathers sufficient information to compensate for sensor errors across different working conditions. This dimensional expansion enables the calibration to be robust against variations in working surface geometry while still being executable with simple ground-based reference.
2Measurement precision
If external measuring devices like total stations are used to improve calibration accuracy, then posture computation precision improves, but the device complexity and operational requirements increase
Solution Approach 1:
The construction machine performs calibration using its own onboard sensors and控制系统, without requiring external measuring devices. The system utilizes the posture information sensors already installed on the machine to detect positions at multiple calibration points, and the onboard computer calculates correction values autonomously. This self-service approach eliminates the need for external equipment like total stations, reducing device complexity and operational requirements while maintaining calibration accuracy through multi-point measurement.
Solution Approach 2:
The onboard posture information sensors and computer system serve dual functions: they perform both the calibration process and the ongoing posture computation during work operations. By making the calibration system universal and integrated into the machine's existing infrastructure, the patent eliminates the need for separate external calibration equipment, thereby reducing overall system complexity while achieving high calibration accuracy.
3Loss of information
If multiple sensors are installed to improve posture detection capability, then the system can capture more comprehensive posture information, but inherent sensor errors increase and reduce computation precision
Solution Approach 1:
The system uses the multi-point calibration data to compute correction values that are fed back to adjust sensor readings during work operations. By establishing the relationship between actual positions (determined through multi-point calibration) and sensor readings at each calibration point, the system generates correction values that compensate for inherent sensor errors. This feedback mechanism allows multiple sensors to work together effectively, maintaining posture information completeness while correcting for individual sensor inaccuracies.
Solution Approach 2:
The calibration process changes the parameters of the sensor system by determining correction values for each sensor based on multi-point measurements. Instead of relying on the raw, potentially erroneous sensor parameters, the system transforms these parameters by applying computed corrections that account for mounting errors and inherent sensor inaccuracies. This parameter transformation enables the multi-sensor system to achieve high precision posture computation despite individual sensor errors.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A hydraulic excavator 100 includes: a multijoint type front implement 1 that is configured by coupling a plurality of driven members 4 to 6 including a bucket 6; inertial measurement units 14 to 16 that detect posture information about the plurality of driven members 4 to 6; and a calibration value computing section 153 that computes calibration parameters used in calibration of detection results of the inertial measurement units 14 to 16; and a work position computing section 154 that computes a relative position of the bucket 6 to the machine body on the basis of the detection results of the inertial measurement units 14 to 16 and the computation result of the calibration value computing section 153, and the calibration value computing section 153 computes the calibration parameters on the basis of the detection results of the inertial measurement units 14 to 16 in a plurality of postures of the front implement 1 in which a reference point set on any of the plurality of driven members 4 to 6 in advance matches a reference position, which differ in a posture of at least one of the plurality of driven members 4 to 6, and the number of which corresponds to the number of the driven members 4 to 6. Thus, highly precisely computation of the posture of the work implement is possible with a simpler configuration.