Physics-Informed Passive Motion Paradigm for Parallel Robots: A High-Precision Motor-Primitives Framework
Fuli Wang, Fazair Nizar Siraj, Windo Hutabarat, Ashutosh Tiwari
AI summary
Problem
Parallel robots struggle with motion generation due to highly nonlinear kinematics and the absence of intuitive motor primitives, while existing control methods rely on computationally heavy numerical solvers or physically uninterpretable neural networks.
Approach
The framework adapts the Passive Motion Paradigm by embedding rigid-body kinematic constraints directly into a neural network, enabling real-time, auto-differentiated Jacobian estimation and force-field-driven motion without iterative solvers.
Key results
- Sub-0.0001 mm geometric consistency errors across 20 test cases on Stewart and Delta platforms
- Real-time, stable Jacobian estimation via auto-differentiation without iterative numerical solvers
- Successful adaptation to distinct parallel mechanisms using self-supervised training
- Physically interpretable, modular control pipeline that generalizes across varying kinematic structures
Why it matters
Provides a robust, high-precision control alternative for parallel robots in manufacturing and assembly by bridging data-driven learning with physical kinematic constraints.
Abstract
Complex embodied systems, whether biological or robotic, must continuously generate goal-directed behaviors while preserving coherence between motor intention and physical feasibility. In parallel robots, this link between intention and me- chanics becomes particularly challenging due to their nonlinear, over-constrained kinematics and the absence of intuitive motor primitives. This letter introduces a passive motion paradigm for parallel robots using self-supervised physics-informed neural networks, which reformulates motion generation as the dynamic unfolding of motor primitives driven by attractor fields in actua- tor space. Unlike traditional forward or optimization-based for- mulations, the framework integrates analytical kinematics with neural fields to ensure both physical consistency and adaptive motion generation. The method estimates the Jacobian matrix as a physically constrained neural field, merging analytical structure with data-driven learning to achieve robust and interpretable be- havior without relying on iterative numerical solvers. Theoretical analysis, simulations, and physical experiments demonstrate the framework’s accuracy, stability, and adaptability across different parallel mechanisms.