NeuRoLab

The project leader is conducting research on the dynamics of multisensory integration (biological sensor fusion, sensory integration) in zebrafish within the scope of the TUBITAK 3501 program, under project number 120E054. The research focuses on understanding how the information from the zebrafish's visual and lateral line organs is weighted and filtered within the central nervous system. Data-driven system identification studies are being conducted to investigate this process. Since these studies are hypothesis-driven, the performance of the resulting models can only be evaluated through their data prediction capabilities. While this validation method is successful for simple superposition-type models, for more complex sensory integration methods such as Kalman Filter (KF), Extended Kalman Filter (EKF), and Bayesian filters, which are likely to emerge in biological systems, this validation method is insufficient to draw definitive conclusions regarding the underlying system dynamics. The aim of this project, which is entirely independent of the ongoing 120E054 project, is to validate the system identification methods used in the project through a physical experimental setup. This will allow for a preliminary evaluation of the frequency responses resulting from the complex filter dynamics (KF, EKF, Bayes), and the system identification methods will be tested on this experimental platform before being applied to the zebrafish. The methods will be refined and finalized based on the performance outcomes of these tests. The urgency of the project can be outlined under the following headings: Animal experiments in the 120E054 project have already begun, and in the coming months, multisensory integration tests and system identification studies will be carried out. Therefore, the proposed experimental setup needs to be quickly prepared and ready for testing. The project team aims to have the system ready within two months, leveraging their experience. Following this, the validation tests and animal experiments will proceed in parallel with the 120E054 project. The project team has already developed the necessary software and electronic infrastructure for the project using Robot Operating System (ROS) on Nvidia Jetson NX boards in the 120E054 project. With this experience fresh, setting up such an experimental setup will be relatively easy. Doing the same work with a new team would be more costly, both in terms of time and financial resources.

In the future, the project leader will propose a new project to transfer the multisensory integration and dynamic sensory weighting methods derived from fish to robotics and unmanned vehicle technologies. In this regard, this project will serve as an important validation tool in the transition from biology to robotic systems. Completing this project quickly will enable the project leader to achieve preliminary results in a robotic system for the next TUBITAK 1001 call.