IBACS Summer Fellow Martin Flament Fultot, will be giving his doctoral dissertation defense on Wednesday, July 22nd at 12:00 PM in a Virtual Conference Room. A copy of the dissertation entitled “Tensegrity and Recurrent Neural Networks: Towards an ecological model of postural coordination” is available from the Graduate Program Office (psychgrad@uconn.edu).
Virtual Meeting Details:
Link: https://zoom.us/j/5610742546?pwd=a3crZURsd2pMN2M4MWFPd0hWeFNCdz09
Meeting ID: 561 074 2546
Password: 1ThVCA
OR
Join by phone: +1 302-273-8917 (PIN: 736590208)
Title of dissertation: Tensegrity and Recurrent Neural Networks: Towards an ecological model of postural coordination
Major Advisor: James Dixon
Abstract: Tensegrity systems have been proposed as both the medium of haptic perception and the functional architecture of motor coordination in animals. However, a full working model integrating those two aspects with some form of neural implementation is still lacking.In this dissertation, a basic two-dimensional cross-tensegrity plant was designed and its mechanics simulated. The plant was coupled to a Recurrent Neural Network (RNN), because of the latter’s anticipatory properties and distributed internal dynamics. The model’s task was to maintain postural balance against gravity despite the intrinsically unstable configuration of the tensegrity plant. The RNN took only proprioceptive input about the springs’ lengths and rate of length change and output minimum lengths for each spring which reset their Hookean profiles and modulated their interaction with the plant’s inertial kinetics. An evolutionary optimization algorithm generated four artificial agents capable of coordinating the patterns of spring contractions in order to maintain dynamic equilibrium. A first study assessed quiet standing performance and revealed coordinative patterns between the tensegrity rods akinto humans’ strategy of anti-phase hip-ankle relative phase. The agents showed a mixture of periodic and aperiodic trajectories of their Center of Mass. Moreover, the agents seemed to tune to the anticipatory “time-to-balance” quantity in order to maintain their movements within a region of reversibility. A second study perturbed the system with mechanical platform shifts and sensorimotor degradation. The agents’ response to the mechanical perturbation was robust and most could maintain balance afterwards. Dimensionality analysis of the RNNs’ unit activations revealed a pattern of slight degree of freedom recruitment during the timesteps following the onset of perturbation. In the degradation sub-study, different levels of noise were added to the inputs to the RNN and different levels of weakening gain were applied to the forces generated by the springs to mimic haptic degradation and muscular weakening in elderly humans. As expected, the systems performed less well, falling earlier than without the insults. However, the same systems re-evolved again under the degraded conditions saw significant functional recovery. Overall, the dissertation supports the plausibility of RNN cum tensegrity models of haptic-guided postural coordination in humans.