Localization with Active Particle Filter Networks

approach

Accurate localization is a critical requirement for most robotic tasks. The main body of existing work is focused on passive localization in which the motions of the robot are assumed given, abstracting from their influence on sampling informative observations. While recent work has shown the benefits of learning motions to disambiguate the robot’s poses, these methods are restricted to granular discrete actions and directly depend on the size of the global map. We propose Active Particle Filter Networks (APFN), an approach that only relies on local information for both the likelihood evaluation as well as the decision making. To do so, we couple differentiable particle filters with a reinforcement learning agent that attends to the most relevant parts of the map. The resulting approach inherits the computational benefits of particle filters and can directly act in continuous action spaces while remaining fully differentiable and thereby end-to-end optimizable as well as agnostic to the input modality. We demonstrate the benefits of our approach with extensive experiments in photorealistic indoor environ- ments built from real-world 3D scanned apartments.


How Does It Work?

approach
Figure: Architecture.


Given the robot’s observations, the localization module, instantiated as a PF-net, updates the belief over the current pose of the robot, modeled as a particle distribution. This distribution is then projected into a belief map over the environment. A reinforcement learning agent then attends to the local regions across the most likely hypotheses as well as the raw robot observations and produces actions at to move the robot’s base, which then result in the next sensory observations for the PF-net.

Videos

Code

A software implementation of this project can be found in our GitHub repository for academic usage and is released under the GPLv3 license. For any commercial purpose, please contact the authors.

Publications

Daniel Honerkamp, Suresh Guttikonda and Abhinav Valada,
Active Particle Filter Networks: Efficient Active Localization in Continuous Action Spaces and Large Maps
arXiv preprint arXiv:2209.09646, 2022.

(Pdf) (Bibtex)


Acknowledgements

This work was funded by the European Union’s Horizon 2020 research and innovation program under grant agreement No 871449-OpenDR.

People