A sensor fusion framework for intelligent navigation and control of autonomous robots

University of Turku

Modern robotic systems such as unmanned aerial vehicles (UAV) and autonomous surface vessels (ASV) have witnessed significant advancements, but achieving effective decision-making in dynamically changing environments remains a huge challenge. Conventional methodologies often ignore the dynamic nature of real-world scenarios, resulting in suboptimal performance. Sensor data fusion offers a promising avenue to merge diverse sensor inputs and improve situational awareness. This research seeks to leverage sensor data fusion techniques to create a holistic index that represents the state of the autonomous robot. Coupled with a machine learning-based supervisory algorithm, this index will enable adaptive adjustments of controller parameters, enabling UAVs and USVs to navigate diverse and dynamically changing environments with greater efficiency. Specifically, the focus is on using them for water monitoring and data collection to aid the development of Digital Twins for integrated water resource management.

This research aims to improve the decision-making capabilities of autonomous robotic systems through an innovative framework that integrates sensor data fusion techniques, advanced control schemes, and machine learning algorithms. The project will explore the integration of sensor data from multiple modalities, such as LiDAR and cameras, as well as thermal and inertial sensors, to generate a holistic understanding of the robot’s environment and internal state. Through sophisticated data fusion algorithms, a comprehensive index will be obtained, capturing both intrinsic and extrinsic features, which are crucial for informed decision-making. Moreover, the implementation of a machine learning-based supervisory algorithm will facilitate intelligent adjustments to the controller parameters, thereby ensuring the robot’s robust and adaptive behavior even in dynamically changing environmental conditions. This holistic framework aims to not only increase the autonomy and adaptability of robotic systems, but also promote their seamless integration into real-world applications, ultimately advancing the frontiers of autonomous robotics. We will carry out missions in the Aura and Teno river environments.

We are looking for a doctoral researcher with a blend of independence and collaboration skills, motivated to work in a multinational and interdisciplinary team on projects involving robotic systems and embedded AI. An outstanding master’s degree in mechanical engineering, mechatronics, automation, computer science, or related fields, including master’s students who will complete their master’s degree in 2024 by the date set by the involved doctoral programme at UTU. Essential skills include proficiency in Python and C++ programming. Additionally, we value: – Experience with robotic hardware platforms and sensor systems. – Knowledge in multi-modal sensor data fusion. – Background in machine learning and control engineering. – Familiarity with ROS for robotics programming. Candidates should be prepared to engage in interdisciplinary research, demonstrating the ability to integrate methods and insights across different fields, and leverage collaboration opportunities.

TBA

Department of Computing, Faculty of Technology

Assoc. Prof. Wallace Moreira Bessa, UTU

Prof. Tomi Westerlund, UTU

Dr. Jani Heikkinen, UTU

TBA