<p>In this paper, a real-time nonlinear model predictive control (NMPC) approach is proposed for high-precision waypoint following in mobile robots with chain-based drive systems. The NMPC-based control setup profits from predictive capabilities and adeptness at handling constraints to deal with the challenges of multi-constrained control problems. An extended Kalman filter (EKF) is integrated into the system framework to achieve real-time estimation for unobservable system states and noisy velocity measurements, thereby enhancing the robustness of the feedback loop. The proposed approach is compared with heuristic methods such as pure pursuit (PP), multi-goal stabilization (MGS), classical sliding mode control (SMC), as well as a standard NMPC formulation with terminal cost (T-NMPC) to showcase improvements in waypoint following performance, disturbance rejection, and time-varying constraints. Based on hardware tests, an extensive real-time feasibility analysis is presented, which is essential for assessing the suitability of the approach for execution on the target hardware platform. The NMPC framework demonstrates improved waypoint following, robustness, and error minimization through MATLAB/Simulink simulations with adaptive variation to operating conditions. The robustness of NMPC is further confirmed through extensive experiments on a real chain-based mobile robot platform developed under ROS2 as a software framework, where <InlineEquation ID="IEq1"> <EquationSource Format="TEX">$R^2$</EquationSource> <EquationSource Format="MATHML"><math> <msup> <mi>R</mi> <mn>2</mn> </msup> </math></EquationSource> </InlineEquation> values reached up to <InlineEquation ID="IEq2"> <EquationSource Format="TEX">$68.18\%$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>68.18</mn> <mo>%</mo> </mrow> </math></EquationSource> </InlineEquation> in simulations and <InlineEquation ID="IEq3"> <EquationSource Format="TEX">$62.01\%$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>62.01</mn> <mo>%</mo> </mrow> </math></EquationSource> </InlineEquation> in real-world experiments, compared to <InlineEquation ID="IEq4"> <EquationSource Format="TEX">$51.42\%$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>51.42</mn> <mo>%</mo> </mrow> </math></EquationSource> </InlineEquation> and <InlineEquation ID="IEq5"> <EquationSource Format="TEX">$57.54\%$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>57.54</mn> <mo>%</mo> </mrow> </math></EquationSource> </InlineEquation> obtained with PP. Stress-testing in the real-time feasibility analysis showed a runtime of approximately <InlineEquation ID="IEq6"> <EquationSource Format="TEX">$50\,\text {ms}$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>50</mn> <mspace width="0.166667em" /> <mtext>ms</mtext> </mrow> </math></EquationSource> </InlineEquation> on target hardware, while the nested-NMPC (N-NMPC) variant achieved runtimes of around <InlineEquation ID="IEq7"> <EquationSource Format="TEX">$20\,\text {ms}$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>20</mn> <mspace width="0.166667em" /> <mtext>ms</mtext> </mrow> </math></EquationSource> </InlineEquation> in real-world experiments due to warm-starting of the optimizer. These findings confirm NMPC as a benchmark in model-based approaches for achieving higher accuracy and control in waypoint following of chain-based robots, thereby contributing to future directions of research in this area.</p>

Real-Time Feasible Nonlinear Model Predictive Control for Waypoint Following of a Chain-Based Mobile Robot

期刊论文
发表日期：

2026年4月8日

查看全文

Hamza Hobbani,
Fernando Perez-Peña,
Karsten Schmidt

摘要

In this paper, a real-time nonlinear model predictive control (NMPC) approach is proposed for high-precision waypoint following in mobile robots with chain-based drive systems. The NMPC-based control setup profits from predictive capabilities and adeptness at handling constraints to deal with the challenges of multi-constrained control problems. An extended Kalman filter (EKF) is integrated into the system framework to achieve real-time estimation for unobservable system states and noisy velocity measurements, thereby enhancing the robustness of the feedback loop. The proposed approach is compared with heuristic methods such as pure pursuit (PP), multi-goal stabilization (MGS), classical sliding mode control (SMC), as well as a standard NMPC formulation with terminal cost (T-NMPC) to showcase improvements in waypoint following performance, disturbance rejection, and time-varying constraints. Based on hardware tests, an extensive real-time feasibility analysis is presented, which is essential for assessing the suitability of the approach for execution on the target hardware platform. The NMPC framework demonstrates improved waypoint following, robustness, and error minimization through MATLAB/Simulink simulations with adaptive variation to operating conditions. The robustness of NMPC is further confirmed through extensive experiments on a real chain-based mobile robot platform developed under ROS2 as a software framework, where $R^2$ $R^{2}$ values reached up to $68.18\%$ $68.18 %$ in simulations and $62.01\%$ $62.01 %$ in real-world experiments, compared to $51.42\%$ $51.42 %$ and $57.54\%$ $57.54 %$ obtained with PP. Stress-testing in the real-time feasibility analysis showed a runtime of approximately $50\,\text {ms}$ $50 ms$ on target hardware, while the nested-NMPC (N-NMPC) variant achieved runtimes of around $20\,\text {ms}$ $20 ms$ in real-world experiments due to warm-starting of the optimizer. These findings confirm NMPC as a benchmark in model-based approaches for achieving higher accuracy and control in waypoint following of chain-based robots, thereby contributing to future directions of research in this area.