Thesis proposals

I often and gladly supervise students doing their theses and internships in our research labs, in particular in the one I am the head of, the Evolutionary Robotics and Artificial Life Lab.

Here you can find a list of topics, divided in three categories, partially overlapping. Most of them are suitable for doing a research thesis projects, i.e., something that hopefully will be finalized in a research article to be submitted to an international conference or journal. Most of them are adequate, in terms of expected quality and quantity of effort to devote, to be tackled as master thesis project, i.e., ranging from 15 to 24 CFUs.

To know more about one or few specific topics, please contact me.

Currently available topics # ↰

Evolutionary robotics # ↰

  • Learning techniques for the controllers of simulated modular soft robots: reinforcement learning
  • Auto-assembly of simulated modular soft robots
  • Resolution-agnostic representation for evolution of closed-loop controllers of simulated modular soft robots
    • see Wang, Yuxing, et al. “Curriculum-based co-design of morphology and control of voxel-based soft robots.” The Eleventh International Conference on Learning Representations. 2023.
  • Social/cultural/imitation learning for simulated modular soft robots
    • see Hart, Emma, and Léni K. Le Goff. “Artificial evolution of robot bodies and control: on the interaction between evolution, learning and culture.” Philosophical Transactions of the Royal Society B 377.1843 (2022): 20210117.
  • Hierarchical (module role-based or environmental context-based) policies for simulated modular robots
  • A embodied/situated, structurally plastic (Hebbian-like) neural network for simulated robotic agents
    • see Najarro, Elias, Shyam Sudhakaran, and Sebastian Risi. “Towards self-assembling artificial neural networks through neural developmental programs.” Artificial Life Conference Proceedings 35. Vol. 2023. No. 1. One Rogers Street, Cambridge, MA 02142-1209, USA journals-info@ mit. edu: MIT Press, 2023.
    • Winther Pedersen, Joachim, et al. “Structurally Flexible Neural Networks: Evolving the Building Blocks for General Agents.” arXiv e-prints (2024): arXiv-2404.
    • Ferigo, Andrea, Elia Cunegatti, and Giovanni Iacca. “Neuron-centric Hebbian Learning.” arXiv preprint arXiv:2403.12076 (2024).

Artificial life # ↰

  • Evolutionary optimization of synapsis-wise and reward-driven autoadaptation rules
    • see Plantec, Erwan, et al. “Evolving Self-Assembling Neural Networks: From Spontaneous Activity to Experience-Dependent Learning.” arXiv preprint arXiv:2406.09787 (2024).
    • see Arnold, Solvi, et al. “Breaching the Bottleneck: Evolutionary Transition from Reward-Driven Learning to Reward-Agnostic Domain-Adapted Learning in Neuromodulated Neural Nets.” arXiv preprint arXiv:2404.12631 (2024).
  • Autogenerating neural-networks

Evolutionary computation # ↰

  • Lexicase-like selection for control problems
    • see: Stanton, Adam, and Jared M. Moore. “Lexicase selection for multi-task evolutionary robotics.” Artificial Life 28.4 (2022): 479-498.
  • LLM-based case generation for lexicase selection
  • Map-elites merges lexicase selection (descriptors capture cases)
  • Adaptive crossover for Map-elites
  • Two- or multi-stage quality-diversity evolution with a representation based on anchor solutions
  • Diversity promotion based on neutrality graph
  • Location-based adaptive surrogate fitness (based on Map-elites or CAbEA for providing locations)
  • Adaptive strategy for selection in Map-elites based on ancestry-similarity
  • The relation between traits heritability/locality and their efficacy as Map-elites descriptors
    • see De Carlo, Matteo, et al. “Heritability of morphological and behavioural traits in evolving robots.” Evolutionary Intelligence (2023): 1-17.
    • see Marrero, Alejandro, et al. “Learning Descriptors for Novelty-Search Based Instance Generation via Meta-evolution.” Proceedings of the Genetic and Evolutionary Computation Conference. 2024.
    • see Cully, Antoine. “Autonomous skill discovery with quality-diversity and unsupervised descriptors.” Proceedings of the Genetic and Evolutionary Computation Conference. 2019.
    • see Tarapore, Danesh, et al. “How do different encodings influence the performance of the map-elites algorithm?.” Proceedings of the Genetic and Evolutionary Computation Conference 2016. 2016.
    • see Tarapore, Danesh, and Jean-Baptiste Mouret. “Evolvability signatures of generative encodings: beyond standard performance benchmarks.” Information Sciences 313 (2015): 43-61.

Ongoing topics # ↰

  • Co-evolutionary Map-elites
  • Variable substrate neural cellular automata
  • Map-elites with trajectory-based, instead of point-based, population arrangement
    • see Flageat, Manon, and Antoine Cully. “Uncertain quality-diversity: evaluation methodology and new methods for quality-diversity in uncertain domains.” IEEE Transactions on Evolutionary Computation (2023).
  • Program synthesis using graph-based genetic programming for Petri Nets-like graphs