Learning-Based Approximate Nonlinear Model Predictive Control Motion Cueing

2025-04-01

AI Summary

To address the poor real-time performance and limited high-frequency control capability of nonlinear model predictive control (NMPC) in serial-robot motion simulators, this paper proposes a learning-based motion prompting algorithm. The method uniquely integrates NMPC’s modeling accuracy with the inference efficiency of deep policy networks, establishing a differentiable predictive control framework that explicitly encodes hard constraints on joint-space velocity, acceleration, and position, and realizes end-to-end mapping grounded in the nonlinear dynamics model. Compared to conventional NMPC, our approach achieves a 400× average speedup in computation while preserving kinematic accuracy and constraint satisfaction. Motion prompting quality—measured by RMSE and correlation coefficient—matches state-of-the-art NMPC, and the method demonstrates strong generalization across diverse robotic platforms and simulation environments. It has been successfully deployed on a real-time physics simulation platform.

Technology Category

• Intelligent Robots: Motion and Path Planning
• Search and Optimization: Sampling/Simulation-based Search
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Motion Cueing Algorithms (MCAs) encode the movement of simulated vehicles into movement that can be reproduced with a motion simulator to provide a realistic driving experience within the capabilities of the machine. This paper introduces a novel learning-based MCA for serial robot-based motion simulators. Building on the differentiable predictive control framework, the proposed method merges the advantages of Nonlinear Model Predictive Control (NMPC) - notably nonlinear constraint handling and accurate kinematic modeling - with the computational efficiency of machine learning. By shifting the computational burden to offline training, the new algorithm enables real-time operation at high control rates, thus overcoming the key challenge associated with NMPC-based motion cueing. The proposed MCA incorporates a nonlinear joint-space plant model and a policy network trained to mimic NMPC behavior while accounting for joint acceleration, velocity, and position limits. Simulation experiments across multiple motion cueing scenarios showed that the proposed algorithm performed on par with a state-of-the-art NMPC-based alternative in terms of motion cueing quality as quantified by the RMSE and correlation coefficient with respect to reference signals. However, the proposed algorithm was on average 400 times faster than the NMPC baseline. In addition, the algorithm successfully generalized to unseen operating conditions, including motion cueing scenarios on a different vehicle and real-time physics-based simulations.

Problem

• Develops learning-based MCA for real-time robot motion simulators
• Combines NMPC accuracy with machine learning efficiency
• Ensures high-speed performance and generalization to new scenarios

Innovation

• Learning-based MCA for real-time robot motion simulators
• Combines NMPC accuracy with machine learning efficiency
• Offline-trained policy network mimics NMPC behavior