On Sunday 19 January 2025, Paderborn University celebrated its traditional New Year's reception with around 400 guests from academia, business and politics.
At this event, Alexander Philipp Nowosad received an award for his outstanding thesis entitled "Design and Realisation of an Intra-FPGA ROS 2 Communication Infrastructure for the ReconROS Executor". Congratulations to him!
Nowosad achieved a significant improvement for FPGA-based robotic systems with his master's thesis. Until now, it was necessary to return all communication between components of an FPGA to the main memory or the software when a component was replaced.
Thanks to the work of Alexander Philipp Nowosad, it is now possible to exchange components dynamically on the FPGA without interrupting the existing internal communication. This significantly increases the efficiency of the system and enables flexible adjustments in real time.
Building on this work, his current research focus is in the field of robotics with FPGAs, which is less about the specific communication and exchange of components and more about accelerating certain parts of robotics applications in a targeted manner.
"I am delighted to have Alexander Philipp Nowosad as a research associate and look forward to continuing our collaboration!" says the head of the Computer Engineering group, Prof. Dr. Marco Platzner.
Laudatory speech by Prof. Dr. Marco Platzner
Mr Alexander Philipp Nowosad's master's thesis deals with the use of field-programmable gate arrays (FPGAs) in robotic systems. FPGAs are electronic components that can adapt their hardware to the function to be calculated by means of a programming process. For many applications, this type of flexible hardware allows functions to be implemented with significantly higher performance and greater energy efficiency than is possible with conventional microprocessors. These FPGA implementations are referred to in the literature as hardware accelerators. There are also functions with high computing power requirements in the field of robotics for which FPGAs are very well suited. Examples of this are the processing of sensor signals and methods for the localisation and navigation of robots.
The use of FPGAs as hardware accelerators in robotics is currently a much-discussed topic with an increasing number of scientific publications and the first commercial products. However, the use of this technology is currently still severely limited by the complexity of FPGA programming. Traditionally, FPGA programming requires knowledge of hardware design and advanced capabilities of FPGAs, such as dynamic and partial reconfiguration, where only part of the hardware resources are adapted at runtime, as well as in-depth expert knowledge. This level of hardware programming is in stark contrast to the abstractions of software programming. In the field of robotics, the Robot Operating System (ROS) is now considered the de facto standard. In terms of a software abstraction, ROS models a robotics application using a set of functions, the so-called ROS nodes, which can communicate via different types of communication. On a technical level, ROS is a distributed middleware that is based on a guest operating system.
The new approach that Mr Nowosad is pursuing in his work is the integration of hardware accelerators as so-called hardware threads in ROS-based applications. Threads are well-known and widely used software abstractions in programming, especially in operating systems and embedded systems. Threads also play a central role in ROS, where ROS nodes are implemented as operating system threads. The new approach now makes it possible to implement ROS nodes not only in software, but also in FPGA hardware, whereby the view of both implementation forms of the threads, software or hardware, is identical, thus greatly simplifying the integration of hardware accelerators into the ROS world. However, the fact that several ROS nodes can be dynamically loaded and executed in parallel in FPGAs, which communicate with each other and with the ROS nodes in software, results in a number of new challenges. These include the development of methods for efficient communication between ROS nodes in hardware without having to go through the operating system level and the avoidance of deadlocks in the execution of ROS nodes.
Based on a comprehensive and deep analysis of related work, Mr Nowosad has tackled these challenges. On the one hand, he has made strong conceptual contributions to solving the problems, but he has also taken on the major challenge of practical implementation in software and hardware and the development of a prototype system with an extraordinarily high level of commitment. A quantitative experimental evaluation of the approach is only possible on the basis of a prototype. His developments are of excellent quality and undoubtedly represent an important contribution to the efficient and flexible use of FPGA technology in robotics, both in a scientific context and in terms of practical feasibility.
The written elaboration is written in English, very clearly structured throughout and extremely carefully designed. It therefore reflects the excellent quality of the work. In the presentations of his work, Mr Nowosad was able to explain the motivation, the challenges and his innovative solutions very clearly. I am delighted that Mr Nowosad, an exceptional young scientist, has received this award and would like to congratulate him warmly, together with my best wishes for what is sure to be a successful future.
Prof Dr Marco Platzner
