Nonlinear Model Predictive Control with Single-Shooting Method for Autonomous Personal Mobility Vehicle

2025-09-20

AI Summary

This work addresses pose control and obstacle avoidance for single-passenger electric autonomous transport vehicles (SEATERs). We propose a real-time motion planning method based on nonlinear model predictive control (NMPC). To enhance computational efficiency and robustness, we innovatively integrate single-shooting within the NMPC framework, augmented with odometry feedback for accurate pose tracking and dynamic obstacle avoidance. The approach explicitly incorporates vehicle nonholonomic constraints, actuation limits, and environmental obstacle constraints. System integration and validation are performed in the ROS/Gazebo simulation environment. Results demonstrate stable convergence to the target pose in both obstacle-free and static-obstacle scenarios, satisfying real-time performance (<50 ms per iteration) and safety requirements. Compared to conventional multi-shooting or linearized NMPC approaches, the proposed method achieves superior tracking accuracy, constraint satisfaction, and computational efficiency.

Technology Category

• Intelligent Robots: Motion and Path Planning
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems

Abstract

This paper introduces a proposed control method for autonomous personal mobility vehicles, specifically the Single-passenger Electric Autonomous Transporter (SEATER), using Nonlinear Model Predictive Control (NMPC). The proposed method leverages a single-shooting approach to solve the optimal control problem (OCP) via non-linear programming (NLP). The proposed NMPC is implemented to a non-holonomic vehicle with a differential drive system, using odometry data as localization feedback to guide the vehicle towards its target pose while achieving objectives and adhering to constraints, such as obstacle avoidance. To evaluate the performance of the proposed method, a number of simulations have been conducted in both obstacle-free and static obstacle environments. The SEATER model and testing environment have been developed in the Gazebo Simulation and the NMPC are implemented within the Robot Operating System (ROS) framework. The simulation results demonstrate that the NMPC-based approach successfully controls the vehicle to reach the desired target location while satisfying the imposed constraints. Furthermore, this study highlights the robustness and real-time effectiveness of NMPC with a single-shooting approach for autonomous vehicle control in the evaluated scenarios.

Problem

• Controls autonomous personal mobility vehicles using Nonlinear Model Predictive Control
• Solves optimal control problems via single-shooting nonlinear programming
• Achieves target pose while satisfying obstacle avoidance constraints

Innovation

• Nonlinear Model Predictive Control for autonomous vehicle navigation
• Single-shooting method solving optimal control via nonlinear programming
• Odometry feedback guides vehicle with obstacle avoidance constraints