Mathematics

A Quantum-Inspired Hybrid Artificial Neural Network for Identifying the Dynamic Parameters of Mobile Car-Like Robots

Joslin Numbi, Mehdi Fazilat, Nadjet Zioui

Published Sep 4, 2025• 6 citations

Abstract

Accurate prediction of a robot’s dynamic parameters, including mass and moment of inertia, is essential for adequate motion planning and control in autonomous systems. Traditional methods often depend on manual computation or physics-based modelling, which can be time-consuming and approximate for intricate, real-world environments. Recent advances in machine learning, primarily through artificial neural networks (ANNs), offer profitable alternatives. However, the potential of quantum-inspired models in this context remains largely uncharted. The current research assesses the predictive performance of a classical artificial neural network (CANN) and a quantum-inspired artificial neural network (QANN) in estimating a car-like mobile robot’s mass and moment of inertia. The predictive accurateness of the models was considered by minimizing a cost function, which was characterized as the RMSE between the predicted and actual values. The outcomes indicate that while both models demonstrated commendable performance, QANN consistently surpassed CANN. On average, QANN achieved a 9.7% reduction in training RMSE, decreasing from 0.0031 to 0.0028, and an 84.4% reduction in validation RMSE, dropping from 0.125 to 0.0195 compared to CANN. These enhancements highlight QANN’s singular predictive accuracy and greater capacity for generalization to unseen data. In contrast, CANN displayed overfitting tendencies, especially during the training phase. These findings emphasize the significance of quantum-inspired neural networks in enhancing prediction precision for involved regression tasks. The QANN framework has the potential for wider applications in robotics, including autonomous vehicles, uncrewed aerial vehicles, and intelligent automation systems, where accurate dynamic modelling is necessary.

DOI: 10.3390/math13172856

Related papers

Quantum Machine Learning-based Test Oracle for Autonomous Mobile Robots

Xinyi Wang, Qinghua Xu, Paolo Arcaini +2 more

Robots are increasingly becoming part of our daily lives, interacting with both the environment and humans to perform their tasks. The software of such robots often undergoes upgrades, for example, to add new functionalities, fix bugs, or delete obsolete functionalities. As a result, regression testing of robot software becomes necessary. However, determining the expected correct behavior of robots (i.e., a test oracle) is challenging due to the potentially unknown environments in which the robots must operate. To address this challenge, machine learning (ML)-based test oracles present a viable solution. This paper reports on the development of a test oracle to support regression testing of autonomous mobile robots built by PAL Robotics (Spain), using quantum machine learning (QML), which enables faster training and the construction of more precise test oracles. Specifically, we propose a hybrid framework, QuReBot, that combines both quantum reservoir computing (QRC) and a simple neural network, inspired by residual connection, to predict the expected behavior of a robot. Results show that QRC alone fails to converge in our case, yielding high prediction error. In contrast, QuReBot converges and achieves 15% reduction of prediction error compared to the classical neural network baseline. Finally, we further examine QuReBot under different configurations and offer practical guidance on optimal settings to support future robot software testing.

arXiv:2508.02407IEEE/ACM International Conference on Automated Software Engineering (ASE)Nov 16, 20251 citation

Investigating Quantum Artificial Neural Networks for Singularity Avoidance in Robotic Manipulators

Mehdi Fazilat, Nadjet Zioui

This research explores the application of quantum-inspired neural networks (QNNs) to address the inverse kinematics problem in robotic arms, explicitly focusing on the ABB IRB140, an articulated robot arm with six degrees of freedom. The primary objective is to develop a quantum-inspired activation function for multilayer perceptron (MLP) neural networks. The study evaluates their performance in avoiding singularities by comparing artificial neural networks (ANNs) with quantum neural networks (QNNs). The findings demonstrate that QNNs outperform ANNs in terms of mean absolute error (MAE), achieving a 15.60% lower MAE in the model without singularities and a 16.67% reduction in the Jacobian-based MAE in the model designed to avoid singularities. The study demonstrates that QNNs offer superior accuracy in predicting the robot arm's inverse kinematics, achieving a position error of 1.64 mm and an orientation error of 0.00179 radians while effectively avoiding singularities. These outcomes underscore the potential of quantum-inspired neural networks to enhance robotic arm manipulations' precision, efficiency, and performance.

International Conference on Systems and Control (ICSC)Nov 3, 202411 citations

Hybrid Quanvolutional Regression Neural Network: Quantum Advantage for Near-term Autonomous Machine

Haoyu Wang, Landon Stobaugh, William Stobaugh

Modern progress in computing power provides great capacity to train large machine-learning models for sufficient accuracy demanded by commercial products. Commercial products such as autonomous vehicles are demanding the complexity of neural networks, traditionally increasing model size to achieve high accuracy and low loss to produce intelligent models. Meanwhile, Quantum Computing has recently advanced into the NISQ (Noisy Intermediate-scale Quantum) era, offering a solid foundation for the potential of improving modern computation tasks with quantum algorithms. Previous quantum algorithms (Variational Quantum Circuits) harness the random nature of quantum mechanics to preserve quantum information and improve model training by allowing parallel computing. This paper demonstrates a hybrid methodology implementing a Hybrid Quantum Neural Network (HQNN) to achieve quantum advantage for a convolutional regression model. The resulting hybrid model performance shows an immediate reduction in model complexity and improvement in training efficiency for the autonomous robot, this contrast with the classical control model provides a promising approach for the near-term feasibility of quantum-enhanced ML products for AI industries.

Jan 26, 20242 citations

Quantum neural network-based inverse kinematics of a six-jointed industrial robotic arm

Mehdi Fazilat, Nadjet Zioui

This research examines the potential of quantum-inspired neural networks (QNNs) for solving the inverse kinematics of robotic arms, focusing on the six-degree-of-freedom ABB IRB140 robot. Traditional inverse kinematics approaches face challenges such as non-unique solutions and computational complexity, especially with increasing degrees of freedom. While artificial neural networks (ANNs) have shown promise, they require further improvements, particularly in terms of quantum computing integration. This study introduces a quantum-inspired activation function to multi-layer perceptron neural networks. We compared ANNs and QNNs with and without singularity avoidance, finding that QNNs significantly outperformed ANNs in mean absolute error (MAE), achieving a 15.60% lower MAE in singularity-free models and a 16.67% lower MAE in singularity-avoidance models. The QNNs demonstrated superior precision, with a position error of 1.64 mm and an orientation error of 0.00179 radians when avoiding singularities. These results highlight the potential of QNNs to enhance the precision, efficiency, and performance of robotic arm manipulation. Quantum computing offers advantages including parallelism, quantum entanglement, and quantum annealing, which contribute to the QNNs’ superior performance. Overall, this study represents a practical contribution to robotics and quantum computing, paving the way for future research into applying quantum principles to neural network models for robotics.

Robotics and Autonomous SystemsJul 6, 202516 citations

Hybrid Convolution Regression Model: Quantum Advantage for Near-term Commercial Autonomous Device

Haoyu Wang, Landon Stobaugh, William Stobaugh

Modern progress in computing power provides great capacity to train large machine-learning models for sufficient accuracy needed for commercial products. Commercial products such as autonomous vehicles are demanding the growing specificity and complexity of machine learning models, trading off increasing model size for high accuracy and low loss to produce intelligent models. Meanwhile, Quantum Computing has recently advanced into the NISQ (Noisy Intermediate-scale Quantum) era, offering a solid foundation for the potential of improving modern computation tasks with quantum algorithms. Previous quantum algorithms (Variational Quantum Circuits) harness the random nature of quantum mechanics to preserve quantum information and improve model training by allowing parallel computing. This paper demonstrates a hybrid methodology implementing a Hybrid Quantum Neural Network (HQNN) to achieve quantum advantage for a convolution regression model. The resulting hybrid model performance shows an immediate reduction in model complexity training efficiency for the autonomous robot, this contrast with the classical control model provides a promising approach for the near-term feasibility of quantum-enhanced ML products for AI industries.

Jan 10, 20240 citations

Quantum-enhanced Hybrid Convolution Regression Model for Sample Autonomous Vehicles

Haoyu Wang, Landon Stobaugh, William Stobaugh

Commercial demands for products of Artificial Intelligence are undeniably at its golden age in the modern world. Advances in physical hardware have provided enough computation capacity to train large machine-learning models for decent accuracy. Commercial products such as autonomous vehicles are demanding the growing specificity and complexity of machine learning models, trading off increasing model size for high accuracy and low loss to produce intelligent models. Meanwhile, Quantum Computing has recently advanced into the NISQ (Noisy Intermediate-scale Quantum) era, offering a solid foundation for the potential of improving modern computation tasks with quantum algorithms. Previous quantum algorithms harness the random nature of quantum mechanics to preserve quantum information and construct the basis for Variational Quantum Circuits (VQC). In this paper, we produce three convolution regression models to predict robot directions for a simplified autonomous-driving task and demonstrate a quantum advantage in our quantum-enhanced hybrid approach as compared to two classical models. Our results in this minimal example reflect the near-term feasibility of hybrid quantum models to enhance modern AI industries, which provides a promising vision to incorporate quantum computing in commercial products.

Dec 13, 20230 citations