Publication detail

SENSORIC FUSION FOR 3D DYNAMIC MAPPING

LIGOCKI, A.

Original Title

SENSORIC FUSION FOR 3D DYNAMIC MAPPING

English Title

SENSORIC FUSION FOR 3D DYNAMIC MAPPING

Type

Ph.D. thesis principal points

Language

en

Original Abstract

This thesis describes my ideas and propositions about the second half of my doctoral studies at the Brno University of Technology under the supervision of prof. Ing. Luděk Žalud, Ph.D. The primary topic of my research is focused on wide range of various sensors fusion into single robust model of the robot surrounding, so this model could be used to achieve higher reliability, precision and safety for robot operation. In my work I am using multiple cameras, IMU, GNSS and LIDARs to understand the robot’s surrounding and to create highly reliable and safety model for later collisions prediction and detection or terrain traversability, etc. In this work I am firstly presenting my dissertation aims, the techniques I want to use to achieve mentioned results, the methods for data filtration and data preprocession in separated pipelines and lately the data fusion into the single robust environment model. After that I am introducing my idea of virtual sensor entities and using artificial intelligence techniques in map building. In the second part I am describing the current state-of-the-art which inspired me, like real-time mapping software, advanced neural network applications, etc., which I would like to base on to reach final result. In the third part I am describing the already realized results, like sensory framework realization or the urban self-driving car dataset recording, which show my progress up to these days.

English abstract

This thesis describes my ideas and propositions about the second half of my doctoral studies at the Brno University of Technology under the supervision of prof. Ing. Luděk Žalud, Ph.D. The primary topic of my research is focused on wide range of various sensors fusion into single robust model of the robot surrounding, so this model could be used to achieve higher reliability, precision and safety for robot operation. In my work I am using multiple cameras, IMU, GNSS and LIDARs to understand the robot’s surrounding and to create highly reliable and safety model for later collisions prediction and detection or terrain traversability, etc. In this work I am firstly presenting my dissertation aims, the techniques I want to use to achieve mentioned results, the methods for data filtration and data preprocession in separated pipelines and lately the data fusion into the single robust environment model. After that I am introducing my idea of virtual sensor entities and using artificial intelligence techniques in map building. In the second part I am describing the current state-of-the-art which inspired me, like real-time mapping software, advanced neural network applications, etc., which I would like to base on to reach final result. In the third part I am describing the already realized results, like sensory framework realization or the urban self-driving car dataset recording, which show my progress up to these days.

Keywords

3D mapping, sensor fusion, virtual sensors, autonomous agent

Released

30.04.2019

Pages count

35

Documents

BibTex


@misc{BUT156705,
  author="Adam {Ligocki}",
  title="SENSORIC FUSION FOR 3D DYNAMIC MAPPING",
  annote="This thesis describes my ideas and propositions about the second half of my doctoral
studies at the Brno University of Technology under the supervision of prof. Ing. Luděk
Žalud, Ph.D. The primary topic of my research is focused on wide range of various sensors
fusion into single robust model of the robot surrounding, so this model could be used to
achieve higher reliability, precision and safety for robot operation. In my work I am using
multiple cameras, IMU, GNSS and LIDARs to understand the robot’s surrounding and
to create highly reliable and safety model for later collisions prediction and detection or
terrain traversability, etc. In this work I am firstly presenting my dissertation aims, the
techniques I want to use to achieve mentioned results, the methods for data filtration and
data preprocession in separated pipelines and lately the data fusion into the single robust
environment model. After that I am introducing my idea of virtual sensor entities and
using artificial intelligence techniques in map building. In the second part I am describing
the current state-of-the-art which inspired me, like real-time mapping software, advanced
neural network applications, etc., which I would like to base on to reach final result. In the
third part I am describing the already realized results, like sensory framework realization
or the urban self-driving car dataset recording, which show my progress up to these days.",
  chapter="156705",
  howpublished="online",
  year="2019",
  month="april",
  type="Ph.D. thesis principal points"
}