Neuro based Soft Robots
Animals successfully navigate in an array of environments due to behavioral adaptations in motor responses to specific environmental stimuli. Non-learned, stereotyped motor responses that are conserved within a species and that sub-serve a survival advantage are known as Modal Action Patterns (MAPs). Understanding the biological basis of MAPs is aided by the application of machine learning techniques to the quantitative study of animal behavior. This project will combine the study of natural foraging behaviors in mice and Drosophila larvae with machine learning techniques to identify evolutionarily conserved action sequences that lead to foraging success while allowing animals to actively avoid harmful or neutral stimuli. Students will employ unsupervised machine learning techniques to additionally extract novel motor sequences that predict foraging success in different animals. In the second phase of the project, students will wire control systems that drive similar stimulus evoked motor action sequences to successfully reach specific targets.
Collaborators: Dr. Wanliang Shan, Dr. Yantao Shen, Dr. Jennifer Hoy, Dr. Chris Feldman, Dr. Matteo Aureli, Dr. hao Xu, Dr. Jun Zhang, Dr. Pradeep Menezes
Funding: NSF-REU (Shan (PI), Mathew (participating faculty); $491,713 (2019-2022))