2020 BRN Discussion, page-849

The following comes Mr. Soumyadeep Choudhury’s web site under the heading Research which can be accessed from the link in my earlier post below:
SOUMYADEEP CHOUDHURYGRADUATE RESEARCHER, THE UNIVERSITY OF TEXAS AT DALLASCURRENT RESEARCH WORK snn_navigation.mp4Monitoring the effect of Pre-Synaptic and Post-Synaptic SpikesReinforcement based Robot Navigation using Spiking Neural Network (SNN) I have been working on robot navigation using Spiking Neural Network where the robot's direction and angle of rotation will be controlled by the spike activities of the neurons.The spikes are generated depending on the input information, learning model and synaptic plasticity among the neuron groups.The reward factor is modulated with the eligibility trace in the STDP (Spike-timing-dependent plasticity) functions of each neuron which learns the environmental factors over time as a form of unsupervised learning.The robot has a SNN based trained model to detect the obstacles and the robot's head-direction is controlled by the spiked neuron depending on the target location for navigation. I also have an idea of modulating the reward into the synaptic layer of neuron learning model to improve the navigation accuracy and learning efficiency over time just like a human learns to navigate in a new environment.This research work has been in progress in TCS - Research & Innovation Lab, Kolkata under Mr. Sounak Dey and Dr. Arijit Mukherjee and collaborated with IIT-Madras.