• Corpus ID: 14540975

Combining Q-Learning with Artificial Neural Networks in an Adaptive Light Seeking Robot

@inproceedings{Dini2012CombiningQW,
  title={Combining Q-Learning with Artificial Neural Networks in an Adaptive Light Seeking Robot},
  author={Steve Dini and Mark Serrano},
  year={2012}
}
Q-learning is a reinforcement learning technique that works by learning an action-value function that gives the expected utility of performing a given action in a given state and following a fixed policy thereafter. The basic implementation uses a q-table to store the data. With increasing complexity in the environment and the agent, this approach fails to scale well as the space requirements become prohibitive. In this paper, we investigate an alternative implementation in which we use an… 
Simulation of mobile robot navigation utilizing reinforcement and unsupervised weightless neural network learning algorithm
TLDR
A novel hybrid technique allows the autonomous system to acquire knowledge, learn and record experience thus attaining self-learning state and the simulated mobile robot was able to differentiate various shapes of obstacles such as corners and walls; and create complex control sequences of movements to maneuver through these obstacles.
Applying hybrid reinforcement and unsupervised weightless neural network learning algorithm on autonomous mobile robot navigation
TLDR
A novel hybrid selflearning algorithm which combines reinforcement and unsupervised weightless neural network algorithm learning was applied to an autonomous mobile robot navigation system in simulation and physical world and shows that the simulated and physical robot possesses the ability to self-learn.
Towards continuous control of flippers for a multi-terrain robot using deep reinforcement learning
TLDR
This work is based on the deep deterministic policy gradient (DDPG) algorithm, proven to be very successful in simple simulation environments, and works in an end-to-end fashion in order to control the continuous position of the flippers.
Learning to Seek: Deep Reinforcement Learning for Phototaxis of a Nano Drone in an Obstacle Field
TLDR
This work deploys a deep reinforcement learning model, capable of direct paths even with noisy sensor readings, and demonstrates efficient light seeking by reaching the goal in simulation in 65% fewer steps and with 60% shorter paths, compared to a baseline random walker algorithm.
Learning to Seek: Autonomous Source Seeking with Deep Reinforcement Learning Onboard a Nano Drone Microcontroller
TLDR
To the authors' knowledge, this is the first deep reinforcement learning method that enables source seeking within a highly constrained nano drone demonstrating robust flight behavior.
Reinforcement Learning With Deeping Learning in Pacman
TLDR
A new method to approximate the true value in reinforcement learning by using deep neural network is proposed and the Pacman was simulated by using this method.
Implementasi Q-Learning dan Backpropagation pada Agen yang Memainkan Permainan Flappy Bird
TLDR
Combining Q-learning and backpropagation can reduce agent’s learning time to play Flappy Bird up to 92% and reduce the weights stored in memory up to 94%, compared to regular Q- learning only.
Reinforcement Learning for Fair Dynamic Pricing
TLDR
This paper shows how to solve dynamic pricing by using Reinforcement Learning techniques so that prices are maximized while keeping a balance between revenue and fairness, and demonstrates that RL provides two main features to support fairness in dynamic pricing.
Adaptive reinforcement learning for heterogeneous network selection
Thesis (Ph.D.) -- University of Adelaide, School of Electrical and Electronic Engineering, 2018
Optimisation de la consommation énergétique d'une ligne de métro automatique prenant en compte les aléas de trafic à l'aide d'outils d'intelligence artificielle
En 2014, dans le cadre du Plan Climat, les pays membres de l’Union Europeenne, se sont engages a reduire de pres de 27% leur consommation d’energie. L’un des axes d’etudes concerne l’augmentation de
...
...

References

SHOWING 1-9 OF 9 REFERENCES
Q-Learning in Continuous State and Action Spaces
TLDR
A method suitable for control tasks which require continuous actions, in response to continuous states, is described, which consists of a neureil network coupled with a novel interpolator.
Reinforcement learning for robots using neural networks
TLDR
This dissertation concludes that it is possible to build artificial agents than can acquire complex control policies effectively by reinforcement learning and enable its applications to complex robot-learning problems.
Reinforcement Learning Neural Network to the Problem of Autonomous Mobile Robot Obstacle Avoidance
TLDR
The simulation results show that the simulated robot using the reinforcement learning neural network can enhance its learning ability obviously and can finish the given task in a complex environment.
Q-Learning with Recurrent Neural Networks as a Controller for the Inverted Pendulum Problem
TLDR
This dissertation aims to provide a history of electronics engineering in Turkey and its applications in the aerospace industry from 1989 to 2000.
Neural Networks and Q-Learning for Robotics
TLDR
A Behavior-Based approach to Supervised Learning of a Behavior and a non-explicit Model for Reinforcement Function Design for Cooperative Robotics.
Building a Light-seeking Robot with Q-learning, http://www.informit.com/articles
  • Building a Light-seeking Robot with Q-learning, http://www.informit.com/articles
  • 2002
Building a Light-seeking Robot with Q-learning
  • http://www.informit.com/articles, April
  • 2002
Q-learning with Look-up Table and Recurrent Neural Networks as a Controller for the Inverted Pendulum Problem
  • Q-learning with Look-up Table and Recurrent Neural Networks as a Controller for the Inverted Pendulum Problem