Many organisms collectively gather resources for survival. When a forager is alone, the efficiency of the foraging process depends heavily on the search strategy and the distribution of resources. However, when part of a collective, the individual should adapt its search strategy based on the available information. Through (local) communication, a collective can apply interaction between individuals to potentially improve the foraging efficiency greatly. Hence, the individual dynamics can be adapted to benefit the collective. In fact, such systems are well documented in nature. For example, an individual can choose to transfer information over consuming the resources itself, such that the collective can benefit (see figure). However, due to limitations to both the communication radius and the cognitive capabilities of individuals, there is a dire need to investigate what type of rules a collective system needs to adhere to in order to increase the efficiency of the foraging task.
In this specific thesis, the student will explore current advances in (optimal) foraging for individuals and extrapolate this towards a collective system, investigating what constitutes a potential (collective) memory model such that the search efficiency of the foraging process benefits from having multiple individuals working together. We have developed a model of individual foraging, which efficiency remembers target locations by learning a spatial distribution which can be sampled. This model is currently implemented to optimize individual behavior, however when a collective is considered it needs to be extended. You will take this model as an initial starting position and aim to extend the model towards multiple individuals that work together. Note that the current model can serve as a template, but there are no restriction to the actual model that will be used. You will draw conclusions based on simulations of foraging models.
You will aim to show that there exist memory models and information transfer mechanisms that are realistic and can be applied to both explain observations in nature as well as be introduced to swarm robot systems that need to execute a foraging task. The ultimate goal is to gain more understanding into collective learning and how the dynamics of the individuals can be tailored to serve the collective.
- Investigate the use of memory models in collective foraging tasks.
- Investigate different interaction strategies that may lead to optimal foraging behaviors
- Gain more understanding into collective learning and how the dynamics of the individuals can be tailored to serve the collective.
The work in this master thesis entails:
- Research and verify the state-of-the-art on individual search strategies in specific environments
- Extend available models to collective systems, with the aim of improving the search efficiency further than is possible with a group of non-cooperative individuals
- Investigate memory models and information transfer mechanisms that are realistic and can be applied to explain observations in nature or to be introduced to robot systems that need to execute a foraging task.
- Hybrid foraging in patchy environments using spatial memoryJ Nauta, Y Khaluf, P Simoens, Journal of the Royal Society Interface 17 (166), 20200026
- Enhanced foraging in robot swarms using collective Lévy walksJ Nauta, S Van Havermaet, P Simoens, Y Khaluf, ECAI2020, the 24th European Conference on Artificial Intelligence 325
- Courses: Programming in Python (INF-22306)
- Required skills/knowledge: Computational modelling, data analytics, interest about collective systems, bio-inspired systems.
Key words: Agent-based modelling, (Agent) Simulation Modelling, Collective systems, Foraging
Yara Khaluf (email@example.com)