Stein Variational Probabilistic Roadmaps

@article{Lambert2022SteinVP,
  title={Stein Variational Probabilistic Roadmaps},
  author={Alexander Lambert and Brian Hou and Rosario Scalise and Siddhartha S. Srinivasa and Byron Boots},
  journal={2022 International Conference on Robotics and Automation (ICRA)},
  year={2022},
  pages={11094-11101}
}
Efficient and reliable generation of global path plans are necessary for safe execution and deployment of autonomous systems. In order to generate planning graphs which adequately resolve the topology of a given environment, many sampling-based motion planners resort to coarse, heuristically-driven strategies which often fail to generalize to new and varied surroundings. Further, many of these approaches are not designed to contend with partial-observability. We posit that such uncertainty in… 
1 Citations

Figures and Tables from this paper

A Gaussian variational inference approach to motion planning

The proposed Gaussian variational inference frame- work for the motion planning problem achieves collision avoidance with smooth trajectories, and brings robustness to the deterministic baseline results, especially in challenging environments and tasks.

References

SHOWING 1-10 OF 34 REFERENCES

Bayesian Hilbert Maps for Dynamic Continuous Occupancy Mapping

A variational Bayesian approach to Hilbert mapping, thus eliminating the regularization term typically adjusted heuristically and extended to learn long-term occupancy maps in dynamic environments in a sequential fashion, demonstrating the power of kernel methods to capture abstract nonlinear patterns and Bayesian learning to construct sophisticated models.

A kernel method for the twosample problem

  • Advances in Neural Information Processing Systems, 2008.
  • 2008

Entropy Regularized Motion Planning via Stein Variational Inference

A sampling strategy based on variational inference to generate distributions of feasible, lowcost trajectories for high-dof motion planning tasks, which includes a distributed, particle-based motion planning algorithm which leverages a structured graphical representations for inference over multi-modal posterior distributions.

Roadmap Learning for Probabilistic Occupancy Maps With Topology-Informed Growing Neural Gas

This work proposes the topology-informed growing neural gas algorithm that leverages estimates of probabilistic topological structures computed using persistent homology theory to focus the roadmap learning on challenging regions of the environment that have not yet been learned correctly.

Dual Online Stein Variational Inference for Control and Dynamics

An implicit variational inference algorithm able to estimate distributions over model parameters and control inputs on-the-fly on the fly is devised, enabling the approximation of complex multi-modal posterior distributions, typically occurring in challenging and realistic robot navigation tasks.

Stein Variational Model Predictive Control

This paper proposes a Stein variational gradient descent method to estimate the posterior directly over control parameters, given a cost function and observed state trajectories and shows that this framework leads to successful planning in challenging, non-convex optimal control problems.

Stein Variational Gradient Descent With Matrix-Valued Kernels

This work enhances SVGD by leveraging preconditioning matrices, such as the Hessian and Fisher information matrix, to incorporate geometric information into SVGD updates, and presents a generalization of SVGD that replaces the scalar-valued kernels in vanilla SVGD with more general matrix- valued kernels.

LEGO: Leveraging Experience in Roadmap Generation for Sampling-Based Planning

An algorithm LEGO is presented that addresses the problem of leveraging prior experience to generate roadmaps in sampling-based motion planning by training the CVAE with target samples that satisfy two important criteria.

Motion Planning Networks: Bridging the Gap Between Learning-Based and Classical Motion Planners

This article describes motion planning networks (MPNet), a computationally efficient, learning-based neural planner for solving motion planning problems, and shows that worst-case theoretical guarantees can be proven if this neural network strategy is merged with classical sample-based planners in a hybrid approach.

Learned Critical Probabilistic Roadmaps for Robotic Motion Planning

This work proposes a general method to identify critical states via graph-theoretic techniques and learn to predict criticality from only local environment features through global connections within a hierarchical graph, termed Critical Probabilistic Roadmaps.