• Corpus ID: 219792208

Variational State-Space Models for Localisation and Dense 3D Mapping in 6 DoF

@article{Mirchev2021VariationalSM,
  title={Variational State-Space Models for Localisation and Dense 3D Mapping in 6 DoF},
  author={Atanas Mirchev and Baris Kayalibay and Patrick van der Smagt and Justin Bayer},
  journal={ArXiv},
  year={2021},
  volume={abs/2006.10178}
}
We solve the problem of 6-DoF localisation and 3D dense reconstruction in spatial environments as approximate Bayesian inference in a deep generative approach which combines learned with engineered models. This principled treatment of uncertainty and probabilistic inference overcomes the shortcoming of current state-of-the-art solutions to rely on heavily engineered, heterogeneous pipelines. Variational inference enables us to use neural networks for system identification, while a… 
Tracking and Planning with Spatial World Models
TLDR
A method for real-time navigation and tracking with differentiably rendered world models by planning in a learned 3D spatial world model, combined with a pose estimation algorithm previously used in the context of TSDF fusion, but now tailored to the authors' setting and improved to incorporate agent dynamics.
Intrinsic Control of Variational Beliefs in Dynamic Partially-Observed Visual Environments
TLDR
It is argued that a compact and general learning objective is to minimize the entropy of the agent’s state visitation estimated using a latent state-space model, and this agent learns to discover, represent, and exercise control of dynamic objects in a variety of partially-observed environments sensed with visual observations without extrinsic reward.
Information is Power: Intrinsic Control via Information Capture
TLDR
It is argued that a compact and general learning objective is to minimize the entropy of the agent’s state visitation estimated using a latent state-space model, and this agent learns to discover, represent, and exercise control of dynamic objects in a variety of partiallyobserved environments sensed with visual observations without extrinsic reward.

References

SHOWING 1-10 OF 54 REFERENCES
Assuring good style for object-oriented programs
The language-independent Law of Demeter, which encodes the ideas of encapsulation and modularity in an easy-to-follow form for object-oriented programmers, is presented. The law was developed during
Intensive immunosuppression with antithymocyte globulin and cyclosporine as treatment for severe acquired aplastic anemia.
TLDR
The data support the use of a combination immunosuppressive regimen containing both ATG and CsA as first-line therapy for severe aplastic anemia.
Mother-Child Attachment and Social Anxiety Symptoms in Middle Childhood.
VIMO: Simultaneous Visual Inertial Model-Based Odometry and Force Estimation
TLDR
This work proposes a relative motion constraint combining the robot's dynamics and the external force in a preintegrated residual, resulting in a tightly coupled, sliding-window estimator exploiting all correlations among all variables.
An Introduction to Variational Methods for Graphical Models
TLDR
This paper presents a tutorial introduction to the use of variational methods for inference and learning in graphical models (Bayesian networks and Markov random fields), and describes a general framework for generating variational transformations based on convex duality.
Auto-Encoding Variational Bayes
TLDR
A stochastic variational inference and learning algorithm that scales to large datasets and, under some mild differentiability conditions, even works in the intractable case is introduced.
Approximate Bayesian inference in spatial environments
TLDR
The proposed stochastic recurrent model is a deep variational Bayes filter augmented with a latent global variable---similar to an external memory component---representing the spatially structured environment that inherits favourable properties from neural networks, such as differentiability, flexibility and the ability to train components either in isolation or end-to-end.
A Flexible Heterogeneous Multi-Core Architecture
TLDR
This paper proposed the Flexible Heterogeneous Mul-tiCore processor (FMC), the first dynamic heterogeneous multi-core architecture capable of reconfiguring itself to fit application requirements without programmer intervention, to overcome the memory wall for applications with high memory-level parallelism (MLP).
...
...