Adaptive Genomic Evolution of Neural Network Topologies (AGENT) for State-to-Action Mapping in Autonomous Agents

Behjat, Amir; Chidambaran, Sharat; Chowdhury, Souma

doi:10.48550/arxiv.1903.07107

Cited by 1 publication

(3 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Based on our observations in, 16 using smaller reaction time leads to a more controllable maneuver; these maneuvers are not necessarily energy optimal. Since the focus of the current paper is not about energy optimality but rather the detection quality which might be more useful in practice, it is possible to fix the reaction time to the smallest practical value.…”

Section: A Optimization Problem Definitionmentioning

confidence: 93%

“…A neuroevolution approach based on the AGENT architecture 16 is utilized here. This approach is hypothesized to be less data hungry (in terms of number of approach scenarios to train) than a supervised learning approach.…”

Section: B Research Objectives: Reciprocal Collision Avoidancementioning

confidence: 99%

“…22,23 Further description of our neuroevolution method called AGENT can be found in. 16 Neuroevolution uses an evolutionary algorithm to optimize the topology and weights of a neural network, and is typically used for solving problems that can be posed as reinforcement learning; 24 however unlike RL, neuroevolution is significantly more amenable to parallel deployment and escaping local minima, both crucial to expensive offline learning investments.…”

Section: B Neuroevolution Processmentioning

confidence: 99%

See 2 more Smart Citations

Training Detection-Range-Frugal Cooperative Collision Avoidance Models for Quadcopters via Neuroevolution

Behjat

Gabani

Chowdhury

2019

AIAA Aviation 2019 Forum

Self Cite

View full text Add to dashboard Cite

Cooperative autonomous approaches to avoiding collisions among small Unmanned Aerial Vehicles (UAVs) is central to safe integration of UAVs within the civilian airspace. One potential online cooperative approach is the concept of reciprocal actions, where both UAVs take pre-trained mutually coherent actions that do not require active online coordination (thereby avoiding the computational burden and risk associated with it). This paper presents a learning based approach to train such reciprocal maneuvers. Neuroevolution, which uses evolutionary algorithms to simultaneously optimize the topology and weights of neural networks, is used as the learning method -which operates over a set of sample approach scenarios. Unlike most existing work (that minimize travel distance, energy or risk), the training process here focuses on the objective of minimizing the required detection range; this has important practical implications w.r.t. alleviating the dependency on sophisticated sensing and their reliability under various environments. A specialized design of experiments and line search is used to identify the minimum detection range for each sample scenarios. In order to allow an efficient training process, a classifier is used to discard actions (without simulating them) where the controller would fail. The model obtained via neuroevolution is observed to generalize well to (i.e., successful collision avoidance over) unseen approach scenarios.

show abstract

Section: A Optimization Problem Definitionmentioning

confidence: 93%

Section: B Research Objectives: Reciprocal Collision Avoidancementioning

confidence: 99%