Smart Aquaculture

In our research, We propose to combine these advanced techniques for fish activity recognition taking into consideration the limitations related to water clarity, turbulences … etc. From the control point of view, we will track the fish using motion modeling-based techniques and classify the fish activity based on its trajectory and its spatiotemporal speed variation. Overall, we aim to learn fish behavior, which will be used as an additional input to the control system design.

Our proposed smart aquaculture system focuses on the following points:

1. Studying the effect of the the environmental conditions on fish health and growth using Reinforcement learning. ( GitHub)

2. Optimal feeding control using MPC controller.( GitHub)

3. Industrial application: Fish health assessment using computer vision . ( GitHub)

   

NB: The project was motivated by the innovative idea of the ( Aquash Startup team)

This reseach project (parts 1 and 2) will be led by a team of three professors. Prof. Laleg’s team has significant expertise in developing control, optimization and monitoring methods. She has been working on simulations and lab experimental testing of her algorithms. Prof. Ghanem’s team is particularly well suited to performing the computer vision tasks. Prof. Berumen’s team will provide their expertise in fish behavior and will help in collecting data. In particular, they will work on labelling the fish behavior and will support the experimental setup with their ongoing aquaria work in KAUST’s marine core lab ( CMRCL).