Guest Talk on Robot Navigation Systems

5 May 2026

Photo: UHH Knowledge Technology

In our Oberseminar on 28^th April 2026, we had a guest talk by Dr. Younès Raoui, Associate Professor at the Faculty of Sciences, Mohammed V University in Rabat, Morocco. Dr. Raoui specializes in neuro-robotics, autonomous navigation and mapping, and nonlinear control for robots. Here are the details of his talk:

An Overview of Neural and Bio-inspired Robot Navigation Systems with Selected Contributions

The bio-inspired robot navigation has become an important field in robotics that uses advances in cognitive science to develop algorithms for planning and localization. The development of foundation models has brought navigation closer to human capabilities, enabling systems to follow textual and image-based instructions.
In this seminar, we provide an introduction to robot navigation and delve into the bio-inspired approach. Neural Object SLAM is a new architecture that creates a semantic map of objects and trains place cell and grid cell networks using a continuous attractor network. It encodes the orientation of the robot with a dynamic neural field that provides an estimation of the robot’s angular velocity. Furthermore, loop closure allows correction of the estimated trajectory of the robot.
A Hebbian learning network is trained during exploration to establish connections between grid cells and familiar places. This approach can be further improved by employing event cameras that train grid cells in the form of an LSTM network in different environments, and then use reinforcement learning to compute the robot commands. Finally, large language models (LLMs) and vision-language models (VLMs) can be combined with Neural Object SLAM to create a bio- inspired navigation system based on foundation models.

Our joint work from ICANN'22 can be found here.