Lie Group Forced Variational Integrator Networks for Learning and Control of Robot Systems
This work addresses the need for efficient and generalizable models in robot control, particularly for autonomous mobile robots, by incorporating physics priors into deep learning, though it is incremental as it builds on existing structure-preserving methods.
The paper tackles the problem of learning accurate robot dynamics models for safe control by introducing LieFVIN, a structure-preserving deep learning architecture that learns controlled Lagrangian or Hamiltonian dynamics on Lie groups from data, enabling accurate and fast predictions without numerical integrators or neural ODEs.
Incorporating prior knowledge of physics laws and structural properties of dynamical systems into the design of deep learning architectures has proven to be a powerful technique for improving their computational efficiency and generalization capacity. Learning accurate models of robot dynamics is critical for safe and stable control. Autonomous mobile robots, including wheeled, aerial, and underwater vehicles, can be modeled as controlled Lagrangian or Hamiltonian rigid-body systems evolving on matrix Lie groups. In this paper, we introduce a new structure-preserving deep learning architecture, the Lie group Forced Variational Integrator Network (LieFVIN), capable of learning controlled Lagrangian or Hamiltonian dynamics on Lie groups, either from position-velocity or position-only data. By design, LieFVINs preserve both the Lie group structure on which the dynamics evolve and the symplectic structure underlying the Hamiltonian or Lagrangian systems of interest. The proposed architecture learns surrogate discrete-time flow maps allowing accurate and fast prediction without numerical-integrator, neural-ODE, or adjoint techniques, which are needed for vector fields. Furthermore, the learnt discrete-time dynamics can be utilized with computationally scalable discrete-time (optimal) control strategies.