Data-Driven Real-Time Magnetic Tracking Applied to Myokinetic Interfaces

Pertuz, Sergio A.; Gherardini, Marta; Santos, Gabriel V. P.; Muñoz, Daniel M.; Ayala, Helon Vicente Vicente Hultmann Hultmann; Cipriani, Christian

doi:10.1109/tbcas.2022.3161133

Cited by 6 publications

(13 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In particular, the multisine acquisition was used for model training and the ramp acquisition was used for model testing. More details of the experimental setup can be obtained in [11].…”

Section: Modelsmentioning

confidence: 99%

“…To estimate the displacement of the five magnets, a modified version of the linear and the RBFNN regression models reported in [11] were used. The models were adapted to address the dimensionality reduction by receiving the principal components that contain meaningful information from the magnetic sensors.…”

Section: Modelsmentioning

confidence: 99%

“…After training the ML models, they were implemented in an FPGA device xc7z020-1clg400c (PYNQ-Z1 development board). We implemented the hardware architecture of the Linear and RBFNN predictor models using our vLinRgen and vRBFgen VHDL code generator tools reported in [11] and [27]. The Vivado HLS tool was used to obtain the VHDL code of the MLPNN model.…”

Section: Physical Implementation Of the Dpr-solutionmentioning

confidence: 99%

“…To estimate the magnet's displacements, two main strategies can be adopted: (a) to solve the inverse problem of magnetostatics, which commonly is addressed by optimization solvers [9] and requires numerical approximations as reported in [6] and [10]; (b) to use data-driven methods and machine learning (ML) models for mapping experimental magnetic measurements to the magnet's displacements [11].…”

mentioning

confidence: 99%

“…The reconfigurable capabilities of the solution allow the adaptation of the hardware to the most reliable model predictor for each magnet. (iii) To reduce the cost of the computational unit as an alternative to the solution in [11].…”

mentioning

confidence: 99%

See 4 more Smart Citations

Dynamic Reconfiguration for Multi-Magnet Tracking in Myokinetic Prosthetic Interfaces

Pertuz,

Mendes,

Gherardini

et al. 2024

Preprint

View full text Add to dashboard Cite

Recently myokinetic interfaces have been proposed to exploit magnet tracking for controlling bionic prostheses. This interface derives information about muscle contractions from permanent magnets implanted into the amputee’s forearm muscles. Machine learning models have been mapped on Field Programmable Gate Arrays (FPGAs) to track a single magnet, achieving good precision and computational efficiency, but consuming a large area and hardware resources. To track several magnets, here we propose a novel solution based on dynamic partial reconfiguration, switching three prediction models: a linear regressor, a radial basis function neural network, and a multi-layer perceptron neural network. A system with five magnets and 128 magnetic sensor inputs was used and experimental data were collected to train a system with five hardware predictors. To reduce the complexity of the models, we applied principal component analysis, ranking by correlation the number of inputs of each model. This run-time reconfigurable solution allows the circuits to be reconfigured in order to select the most reliable predictor model for each magnet while the rest of the circuit continues to operate extracting the most significant information from the captured signals. Thus, the proposed solution remarkably reduces the hardware occupation and improves the computational efficiency compared to previous solutions.

show abstract

“…In particular, the multisine acquisition was used for model training and the ramp acquisition was used for model testing. More details of the experimental setup can be obtained in [11].…”

Section: Modelsmentioning

confidence: 99%

Section: Modelsmentioning

confidence: 99%