JOB DESCRIPTION

Autonomous robots promise to transform our economy and society, thanks to the quick progress in artificial intelligence. However, state-of-the-art AI requires computing devices that are too heavy and energy-consuming. This makes robots heavy and dangerous, and hence unsuitable for many applications.
In the NWO VICI project “Neuromorphic Learning for Advanced Insect-inspired Artificial Intelligence (NL-AI2)”, we will develop a neuromorphic AI that is extremely efficient, as it mimics the sparse and asynchronous processing of the brain. To best demonstrate this technology’s efficiency, the AI will be modelled after insect intelligence and instilled on a resource-restricted drone. This will unlock innumerable new applications and even provides opportunities to learn about nature itself!
Within the project, a PhD position is now available on “Neuromorphic vision for insect-inspired path integration and obstacle avoidance”. Insect navigation heavily relies on path integration, i.e., the integration of velocity estimates over time to determine the position with respect to a starting point. To this end, they likely combine forward models that take into account the effects of their intended actions and sensory feedback from various sensor modalities such as vision. The topic of the PhD is to develop neuromorphic vision algorithms that process (1) events from a neuromorphic camera to estimate ego-motion and depth, and (2) polarization images to estimate the position of the sun. We strive for the development of self-supervised learning algorithms that have the potential to run onboard of an autonomous drone. The research will involve the development of self-supervised learning algorithms for training spiking neural networks. Experiments will initially involve passively gathered data, but will eventually be performed onboard flying drones.
The candidate will be guided by dr. C. De Wagter and prof. dr. G.C.H.E. de Croon at the Micro Air Vehicle laboratory (MAVLab), which is part of the Control and Operations department of the Faculty of Aerospace Engineering at TU Delft. They will have access to state-of-the-art tools and facilities needed to conduct their research. They will have the opportunity to collaborate with other researchers from within the laboratory, as well as on an international level.

JOB REQUIREMENTS

• An excellent degree in computer vision, artificial intelligence, computer science, robotics, or a related subject at Masters level or equivalent
• Good programming skills in python and/or C++
• Fluency in English
• An independent scholar who wants to shape research in the field
• An interest in interdisciplinary and collaborative research

Beneficial:

• Hands-on experience with robot or drone hardware

• Previous research experienceKnowledge of drone autopilots and/or ROS

Please refer the Job description for details