High-accuracy prediction and compensation of industrial robot stiffness deformation

Ye, Congcong; Yang, Jixiang; Ding, Han

doi:10.1016/j.ijmecsci.2022.107638

Cited by 26 publications

(8 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The network architecture is illustrated in Figure 4, and its hidden layers follow the structure outlined in Ye et al (2022), comprising 18, 36, 72, 36, 18, 18 and 9 neurons, respectively. The input and output layers are modified to encompass 12 and 3 neurons for 12-dimensional joint configurations and 3-dimensional relative errors.…”

Section: Calibration-driven Transfer Learning Methods For Relative Er...mentioning

confidence: 99%

“…The term “transfer” refers to the process of acquiring knowledge from one (source) data set and applying it to another related (target) data set (Pan and Yang, 2010). Inspired by simulation-driven transfer (Akhmetzyanov et al , 2020; Liu and Gryllias, 2022; Ye et al , 2022), where simulations are used to generate source data for acquiring transferable knowledge from existing models, we innovatively propose a calibration-driven transfer method, where the calibrated model is not used to directly improve the robot accuracy but is used as a data generator to generate error data between the nominal and calibrated models’ outputs. Pretraining with the generated data enables the network to “learn” from calibration, and subsequent fine-tuning with a small amount of real (measured) data enables the network to further “learn” from the error sources that have been overlooked in calibration, thus achieving higher prediction accuracy.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A method for predicting relative position errors in dual-robot systems via knowledge transfer from geometric and nongeometric calibration

Cao,

Wang,

Guo

et al. 2024

View full text Add to dashboard Cite

Purpose In a dual-robot system, the relative position error is a superposition of errors from each mono-robot, resulting in deteriorated coordination accuracy. This study aims to enhance relative accuracy of the dual-robot system through direct compensation of relative errors. To achieve this, a novel calibration-driven transfer learning method is proposed for relative error prediction in dual-robot systems. Design/methodology/approach A novel local product of exponential (POE) model with minimal parameters is proposed for error modeling. And a two-step method is presented to identify both geometric and nongeometric parameters for the mono-robots. Using the identified parameters, two calibrated models are established and combined as one dual-robot model, generating error data between the nominal and calibrated models’ outputs. Subsequently, the calibration-driven transfer, involving pretraining a neural network with sufficient generated error data and fine-tuning with a small measured data set, is introduced, enabling knowledge transfer and thereby obtaining a high-precision relative error predictor. Findings Experimental validation is conducted, and the results demonstrate that the proposed method has reduced the maximum and average relative errors by 45.1% and 30.6% compared with the calibrated model, yielding the values of 0.594 mm and 0.255 mm, respectively. Originality/value First, the proposed calibration-driven transfer method innovatively adopts the calibrated model as a data generator to address the issue of real data scarcity. It achieves high-accuracy relative error prediction with only a small measured data set, significantly enhancing error compensation efficiency. Second, the proposed local POE model achieves model minimality without the need for complex redundant parameter partitioning operations, ensuring stability and robustness in parameter identification.

show abstract

Section: Calibration-driven Transfer Learning Methods For Relative Er...mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A method for predicting relative position errors in dual-robot systems via knowledge transfer from geometric and nongeometric calibration

Cao,

Wang,

Guo

et al. 2024

View full text Add to dashboard Cite

show abstract

“…In recent years, some scholars adopted the improved whale optimization algorithm to reduce errors [ 13 ]. Other scholars used artificial neural networks to compensate position errors [ 14 , 15 , 16 ], and some scholars used similar methods of neural networks to reduce errors in robots, such as the fuzzy neural network (FNN) [ 17 ] and the dual regression adaptive domain adversarial neural network [ 18 ]. In this paper, the dynamic analysis method is adopted, and the joint driver of the robot is imagined as a spiral spring.…”

Section: Introductionmentioning

confidence: 99%

A Spring Compensation Method for a Low-Cost Biped Robot Based on Whole Body Control

Wang

Kou

et al. 2023

Biomimetics

View full text Add to dashboard Cite

At present, the research and application of biped robots is more and more popular. The popularity of biped robots can be better promoted by improving the motion performance of low-cost biped robots. In this paper, the method of the Linear Quadratic Regulator (LQR) is used to track a robot’s center of mass (COM). At the same time, the whole-body-control method and value function generated in the process of tracking COM are used to construct the quadratic programming (QP) model of a biped robot. Through the above method, the torque feedforward of the robot is obtained in the Drake simulation platform. The torque feedforward information of the robot is transformed into position feedforward information by spring compensation. In this paper, open loop control and spring compensation are used, respectively, to make the robot perform simple actions. Generally, after the compensation method of spring compensation is adopted, the roll angle and pitch angle of the upper body of the robot are closer to 0 after the robot performs an action. However, as the selected motion can introduce more forward and lateral motions, the robot needs more spring clearance compensation to improve performance. For improving the motion performance of a low-cost biped robot, the experimental results show that the spring compensation method is both reasonable and effective.

show abstract

“…In general, robots are more flexible than CNC machines because they have more degrees of freedom and may be used for a variety of tasks, but they are less accurate and precise (especially from the point of view of the trajectory between points) [9]. Robots have inferior geometrical accuracy and precision due to their extended cantilever kinematic structure, which must support the motors, brakes, and reduction gears of each axis, whereas the axes of cartesian CNC machines are stiffer and more robust and less influenced by vibrations [10]. Differences in the kinematic architecture of the two explored configurations lead to variations in trajectory and travel speed management.…”

Section: Introductionmentioning

confidence: 99%

“…In fact, the machining sector is looking to robots for their flexibility, multi-purpose and large-scale capabilities, and reduced cost when compared to CNC machines. Yet, due to the physical interaction between the machined component and the tool, which must apply a force to remove the material, concerns regarding the low precision and stiffness of industrial robots are much more felt [10,11].…”

Section: Introductionmentioning

confidence: 99%

Comparison between Eight-Axis Articulated Robot and Five-Axis CNC Gantry Laser Metal Deposition Machines for Fabricating Large Components

Maffia

Chiappini²,

Maggiani³

et al. 2023

Applied Sciences

View full text Add to dashboard Cite

Laser metal deposition (LMD) is an additive manufacturing (AM) process capable of producing large components for the aerospace and oil and gas industries. This is achieved by mounting the deposition head on a motion system, such as an articulated robot or a gantry computer numerical control (CNC) machine, which can scan large volumes. Articulated robots are more flexible and less expensive than CNC machines, which on the other hand, are more accurate. This study compares two LMD systems with different motion architectures (i.e., an eight-axis articulated robot and a five-axis CNC gantry machine) in producing a large gas turbine axisymmetric component. The same process parameters were applied to both machines. The deposited components show no significant differences in geometry, indicating that the different performances in terms of accuracy of the two machines do not influence the outcome. The findings indicate that LMD can consistently produce large-scale axisymmetric metal components with both types of equipment. For such an application, the user has the option of using an articulated robot when flexibility and cost are essential, such as in a research context, or a CNC machine where ease of programming and process standardization are important elements, such as in an industrial environment.

show abstract

High-accuracy prediction and compensation of industrial robot stiffness deformation

Cited by 26 publications

References 47 publications

A method for predicting relative position errors in dual-robot systems via knowledge transfer from geometric and nongeometric calibration

A method for predicting relative position errors in dual-robot systems via knowledge transfer from geometric and nongeometric calibration

A Spring Compensation Method for a Low-Cost Biped Robot Based on Whole Body Control

Comparison between Eight-Axis Articulated Robot and Five-Axis CNC Gantry Laser Metal Deposition Machines for Fabricating Large Components

Contact Info

Product

Resources

About