Subprojects

M4 Control

Efficient and Accurate State Estimation and Feedback Control under Uncertainties

M4 estimates the hidden state of the running process and steers it, although it is observed only partly and with noise.

Illustration of the control task in a forming simulation of a tape laminate. A friction change half-way through forming lets the laminate go slack. Without control, the flanges wrinkle. With predictive control, the grippers hold back and keep the tension in band. The plot shows the predicted futures and the one applied.

Work packages

Results of the first funding phase, by work package.

  1. M4.1 Stochastic filtering in large, distributed state spaces

    A Newton-flow particle filter formulates the particle update as an ordinary differential equation derived from a generalised Cramér–von Mises distance. The particles keep equal weights, which avoids degeneracy and global resampling, and deterministic Dirac-mixture samples reduce the number of particles needed.

    Publications

  2. M4.2 Accurate sensor models for state estimation

    Bayesian neural networks trained with variational inference, Hamiltonian Monte Carlo and Kalman-filter backends serve as uncertainty-aware sensor models. Local calibration tests based on k-d trees, ball trees, Voronoi regions and kernels reveal where such models are miscalibrated, and new multivariate metrics assess calibration across all outputs.

    Publications

  3. M4.3 Stochastic feedback control of nonlinear distributed processes

    Gradient-free, sampling-based model predictive control, built on the cross-entropy method and model predictive path integral control, handles non-differentiable simulators such as Abaqus. Precomputed deterministic samples cover the solution space with fewer samples and give smoother control inputs. The methods were validated on benchmarks and forming simulations.

    Block diagram of sampling-based model predictive control. A sampling-based optimiser draws control input trajectories, each is shot through the system model, and the trajectory costs return to the optimiser.
    Sampling-based model predictive control allows gradient-free optimisation of control inputs. Sampled control trajectories are used to predict the behaviour of the system (trajectory shooting), which can be parallelised efficiently. Figure: KI-FOR 5339.

    Publications

  4. M4.4 Model-based policy optimisation of nonlinear distributed controllers

    HEPi, the main contribution on the policy-learning side, optimises structure-aware distributed controllers in simulation with a heterogeneous graph and SE(3)-equivariant message passing, improving sample efficiency, final performance and generalisation across object geometries. IGNS adds structure-preserving learned simulators as a basis for future action-conditioned planning and control.

    HEPi architecture. A heterogeneous graph of actuator, object and target nodes is processed by equivariant message passing networks whose outputs are summed into velocity vectors for the actuators.
    Overview of the Heterogeneous Equivariant Policy (HEPi). Equivariant message passing networks process the observation graph and aggregate the actuator, object and target nodes into the action. Figure: Hoang et al., ICLR 2025.

    Publications

About the subproject

M4 is designed to build on the learned models of M2 and the settings found by M3, and to keep the process on track while it runs, where M3 improves it from trial to trial. Because the forming hardware allows only feedforward actuation, its methods were validated on benchmarks and in simulation.

Principal investigators
Researchers
  • Tai Hoang
  • Xiaozhu Luo
  • Markus Walker
Host institute
Intelligent Sensor-Actuator-Systems Laboratory (ISAS), KIT
People in M4

Publications 14