Lehrstuhl für Mechatronik in Maschinenbau und Fahrzeugtechnik (MEC)

Autonomous systems

Motivation

Automation of tasks has been rapidly increasing in the last years. Autonomous systems have the ability to sense, learn and act independently, i.e. without any interference of human operators, even in unforeseen situations. They cover a wide range of areas such as robotics, process technology, smart factories and the omnipresent  self-driving cars. The capabilities of autonomous systems build upon a variety of  methods stemming from different engineering and scientific  disciplines with computer and control sciences as the key ones.

 

Course Content

This course gives an overview of technologies,  methods and algorithms of autonomous systems. The students  learn different concepts including  perception, object tracking, SLAM, planning and control, which represent the core components of  autonomous systems. Additionally, the students  have the opportunity to apply the  concepts in practical examples as part of code demos.

  1. Introduction to autonomous systems, perspectives  and challenges.
  2. Sensing and Actuators: Image based sensors, range based sensors, working principles; actuator technologies
  3. Perception: Camera models; image processing; computer vision.
  4. Deep Neural Networks: CNN/GNN, RNN; supervised learning; object detection, image segmentation; scene understanding.
  5. Estimation: Bayesian inference; Kalman filtering; sensor fusion; multi-object tracking; localization; SLAM.
  6. Planning: Feasible planning, optimal planning, search algorithms, A*, Dijkstra's algorithm, forward / backward search, value iteration.
  7. Decision making and Control: MPC; imitation learning; reinforcement  learning.
  8. Case studies: Sample codes and algorithms for self-driving vehicles and mobile robots in simulation and practice.

 

Literature

  • Thrun, Burgard, Fox, ”Probabilistic Robotics”, MIT Press, 2005.
  • Siegwart, Nourbakhsh, Scaramuzza, ”Introduction to Autonomous Mobile Robots”, MIT Press, 2011.
  • Sutton and Barto, ”Reinforcement learning: An Introduction”, Second Edition, MIT Press, Cambridge, MA, 2018.

Dozenten

Prof. Dr.-Ing. Naim Bajcinca
+49 631/205-3230
Gebäude 42, Raum 262
Sprechstunde: nach Vereinbarung
naim.bajcinca(at)mv.uni-kl.de

Dr. Sandesh Hiremath
Gebäude 65, Raum 420
67663, Kaiserslautern
Phone: +49 631/205-3455
sandesh.hiremath(at)mv.uni-kl.de
 

Lecture

Starting 09.11.2022
KIS entry
 

Exam

Written Exam: 120-150 Min.
Date: 06.03.2023
Time: 11:30 – 14:00
Location: Building. 46-215/46-210/46-110
Credit Points: 4ECTS
KIS entry

 

Prerequisites

Linear Algebra
Probability and Statistics
Fundamentals of robotics

Script

Zum Seitenanfang