1. ICNS Home
  2. Projects
  3. Open PhD Projects
  4. Control Systems Inspired by Insect Central Pattern Generators that can Adapt to Dynamic Environments.

Control Systems Inspired by Insect Central Pattern Generators that can Adapt to Dynamic Environments.

Supervisors:

Primary supervisor Dr Saeed Afshar

Description:

This project will investigate modelling various Central Pattern Generators of insects to apply them for practical control systems. The project is focussed on building control systems targeted towards specific robotic functions that can adapt to dynamic changes in the environment. The project would also involve exploration of adaption and learning mechanisms using the central pattern generators to enhance robustness.

The insect kingdom is full of examples of organisms that are capable of surviving in adverse weather and environmental conditions. Biological systems can adapt to rapid changes in the environments and without any training. Current neural network-based control systems require long periods of simulated training to adapt to various conditions. In this project, we aim to build simple control systems that are inspired by the central pattern generators of various insect nervous systems. These control systems do not require training but are specially designed to adapt to challenging environments and targeted towards a solving a specific control task and not achieve domain general intelligence. The control systems are focussed on simplicity to enable ease of manufacturing and functioning through minimal computational resources.

Figure 3: Hexapod walking robot built performing mistiming in locomotion based on a Central Pattern Generator. (Szadkowski et al. 2021. Reference)

Outcomes:

  • Explore the models of various Central Pattern Generators found in biology and replicate them in simulation to study their behaviour.
  • Investigate control systems that can use the CPG models to perform specialized functions in dynamic environments. Investigate adaptation and learning mechanisms to enable the resulting system to rapidly adopt to changes in the environment.
  • Simulate the control systems in software and benchmark their capacity to adapt to different conditions.
  • Develop working prototype(s) that can be applied to real world scenarios and test their functional capacity under varying conditions.

Eligibility criteria:

Experience with C++, Python, MATLAB, or other equivalent languages for developing and testing the algorithms. Experience with algorithms and strong mathematical background.