Arun Singh presents Reliable Autonomous Navigation with a Single Monocular Camera
On 2026-05-11 10:00:00 at E112, Karlovo náměstí 13, Praha 2
Autonomously navigating between a start and a goal position is the most basic
requirement in a mobile robot. Typically, the primary environment sensor is a
LiDAR that provides the position of the obstacles that need to be avoided.
Moreover, there exist a lot of the off-the-shelf LiDAR based planners that can
provide a decent out of the box performance in even complex environments.
But can we produce a similar level of performance with a single monocular
camera? This is the focus of our research since the last couple of years. Our
primary motivation stems from the fact that there are many robots like
quadrotors which have a small form factor and thus are not ideal to carry heavy
sensors like 3D LiDAR. Moreover, even on ground robots forgoing LiDAR without
compromising performance can save energy and increase the operational cycle.
The talk will cover our recent results on monocular vision based navigation that
combines state-of-the-art deep learning with more classical probabilistic
reasoning and trajectory optimization. I will also show the latest hardware
demonstrations on both ground and quadrotors.
BIO: Arun Kumar Singh leads the AKS Lab
(https://sites.google.com/view/akslab/home) at the Institute of Technology,
where he is also an Associate Professor. At the moment, the lab houses 6 PhD
students, 2 Postdocs, and 1 Research Engineer. He performs fundamental research
in algorithmic foundations of robotics, control systems, and machine learning.
He and his lab have made impactful contributions in developing optimization
algorithms and learning-based robot planning and control in safety-critical
applications through the use of implicit layers embedded neural networks, and
planning under uncertainty. He has a strong publication record in premier
journals such as IEEE Robotics and Automation Letters, IEEE Transactions on
Control Systems Technology, IEEE Transactions on Automation Science and
Engineering, and premier conferences such as ICRA, IROS, CDC. Arun has been PI
for several national and EU projects with the most recent being
XSCAVE(xscave.eu), which seeks to develop deployable AI for heavy machinery.
requirement in a mobile robot. Typically, the primary environment sensor is a
LiDAR that provides the position of the obstacles that need to be avoided.
Moreover, there exist a lot of the off-the-shelf LiDAR based planners that can
provide a decent out of the box performance in even complex environments.
But can we produce a similar level of performance with a single monocular
camera? This is the focus of our research since the last couple of years. Our
primary motivation stems from the fact that there are many robots like
quadrotors which have a small form factor and thus are not ideal to carry heavy
sensors like 3D LiDAR. Moreover, even on ground robots forgoing LiDAR without
compromising performance can save energy and increase the operational cycle.
The talk will cover our recent results on monocular vision based navigation that
combines state-of-the-art deep learning with more classical probabilistic
reasoning and trajectory optimization. I will also show the latest hardware
demonstrations on both ground and quadrotors.
BIO: Arun Kumar Singh leads the AKS Lab
(https://sites.google.com/view/akslab/home) at the Institute of Technology,
where he is also an Associate Professor. At the moment, the lab houses 6 PhD
students, 2 Postdocs, and 1 Research Engineer. He performs fundamental research
in algorithmic foundations of robotics, control systems, and machine learning.
He and his lab have made impactful contributions in developing optimization
algorithms and learning-based robot planning and control in safety-critical
applications through the use of implicit layers embedded neural networks, and
planning under uncertainty. He has a strong publication record in premier
journals such as IEEE Robotics and Automation Letters, IEEE Transactions on
Control Systems Technology, IEEE Transactions on Automation Science and
Engineering, and premier conferences such as ICRA, IROS, CDC. Arun has been PI
for several national and EU projects with the most recent being
XSCAVE(xscave.eu), which seeks to develop deployable AI for heavy machinery.