AUTHOR=Feldotto Benedikt , Lengenfelder Heiko , Röhrbein Florian , Knoll Alois C. 

TITLE=Network Layer Analysis for a RL-Based Robotic Reaching Task

JOURNAL=Frontiers in Robotics and AI

VOLUME=Volume 9 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/robotics-and-ai/articles/10.3389/frobt.2022.799644

DOI=10.3389/frobt.2022.799644

ISSN=2296-9144

ABSTRACT=Recent experiments indicate that pre-training of end-to-end Reinforcement Learning neural networks on general tasks can speed up the training process for specific robotic applications. However, it remains open if these networks form general feature extractors and a hierarchical organization that can be reused as in, e.g., Convolutional Neural Networks. In this paper we analyze the intrinsic neuron activation in networks trained for target reaching of a robot manipulator with increasing numbers of joints. We analyze the individual neuron activity distribution in the network, introduce a pruning algorithm and spot correlations of neuron activity patterns with the resulting dense network representations. We show that the input and output network layers have more distinct neuron activation in contrast to inner layers. Our pruning algorithm reduces  network size significantly, increases the distance of neuron activation while keeping a high performance in training and evaluation. Our results demonstrate that neuron activity can be mapped among networks trained for robots with different complexity. Hereby, robots with small joint differences show higher layer-wise projection accuracy whereas more diverse robotic configurations mostly show projections to the first layer.