I am a PhD student in robotics and computer vision supervised by Patric Jensfelt (started October 2019). My research is part of the Wallenberg AI, Autonomous Systems and Software Program (WASP-AS). The topic of my research is Autonomous Dense Semantic Mapping with Editable 3D Models.
Most dense map representations used in robotics represent the world as monolithic datastructures without explicitly representing objects or properties of the environment. This is fundamentally different from the way humans perceive and interpret the world around them. Objects have associated properties and even from a single view we can easily infer approximate object shapes based on past experience.
Recent advances in deep learning have enabled rapid progression on computer vision tasks such as semantic segmentation and object detection, but lifting these results into 3D space and robust integration into robotic systems is still in its infancy.
I am working on tightly integrating simultaneous localization and mapping with deep generative models (for example, variational autoencoders), semantic/panoptic segmentation and recent advances such as differentiable rendering. The long-term goal is to work towards an object centric world representation in which the world is represented as individual 3D objects with associated properties. Such a representation lends itself naturally to higher level planning, manipulation and efficient storage.
I am happy to supervise motivated bachelor or master students in topics related to the previously mentioned research interests. Feel free to send your CV and transcript of records to . If you already have a topic in mind, please include a project proposal. If not, just state your interests and we can also try to find a potential topic together.
I completed my B.Sc. in 2016 at RWTH Aachen University in Electrical Engineering, Information Technology and Computer Science with a focus on integrated circuit design. From 2016 to 2019 I completed two M.Sc. as part of the T.I.M.E. Double Degree Program focussing on Robotics and Autonomous Systems at KTH and Information and Communication Technology in Aachen.
In the past I have been working as a student at multiple institutes at RWTH Aachen University and as an intern at Deepfield Robotics, Ericsson Research and as a master's thesis student at Robert Bosch GmbH. During this time I have been working predominantly on motion planning, dynamic modeling and augmented reality.
Luigi Palmieri*, Leonard Bruns*, Michael Meurer & Kai O. Arras. Dispertio: Optimal Sampling For Safe Deterministic Motion Planning. IEEE Robotics and Automation Letters, 5(2), 363-369. (also accepted for presentation at IEEE International Conference on Robotics and Automation)
* equal contribution