Autonomous navigation based on a Q-learning algorithm for a robot in a real environment

Strauss, Clément; Sahin, Ferat

doi:10.1109/sysose.2008.4724191

Cited by 9 publications

(3 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Embedded autonomous system face resource-constrained issues [12], [13]; processor speed, storage capacity, run-time memory and other hardware related matters; 4) Autonomous system constitutes of finite states/situations and actions; and expert knowledge is translated into computer program used to activate these actions in order to manipulate the environment and can be classified as followings: a) A system whereby anticipated states are known beforehand, therefore can be generalized by using pre-trained Neural Networks [14]- [17] and expert knowledge was preprogrammed to handle number of actions; b) Or, rather than using pre-programmed expert knowledge, a reinforcement learning algorithm can be applied in order to make the system learn as time progresses [2], [18], [19]. 5) Autonomous system applications developed by Bagnall, Claveau, Nurmaini, Strauss and others [3], [14], [16], [20], [21] demonstrates that both reinforcement learning and weightless neural network algorithm can be successfully applied in autonomous systems which implemented in resource constraint environment;…”

Section: Literature Reviews On Self-learning Andmentioning

confidence: 99%

See 1 more Smart Citation

Simulation of mobile robot navigation utilizing reinforcement and unsupervised weightless neural network learning algorithm

Yusof

Mansor

Baba

2015

2015 IEEE Student Conference on Research and Development (SCOReD)

View full text Add to dashboard Cite

The approach of transforming human expert knowledge into computer program only allow a system to solve foreseen and tested outcomes compared to a system having selflearning capabilities. This paper will summarize and discuss the research, design and implementation of a novel self-learning algorithm which combines: (a) Q-Learning -A reinforcement learning algorithm; and (b) AutoWiSARD -An unsupervised weightless neural network learning algorithm. The self-learning algorithm was implemented in an autonomous mobile robot navigation and obstacle avoidance system in a simulated environment.The AutoWiSARD algorithm identifies, differentiates and classifies the obstacles and the Q-learning algorithm learns and tries to maneuver through these obstacles. This novel hybrid technique allows the autonomous system to acquire knowledge, learn and record experience thus attaining self-learning state. The final result shows 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.

show abstract

Section: Literature Reviews On Self-learning Andmentioning

confidence: 99%

“…The formulated self-learning algorithm employs; a) Unsupervised Weightless Neural Network which handles generalization issues; b) A reinforcement learning algorithm which act as a reward and penalized function by learning the optimal policy from its history of interaction with the environment [2], [3].…”

Section: Introductionmentioning

confidence: 99%

Simulation of mobile robot navigation utilizing reinforcement and unsupervised weightless neural network learning algorithm

Yusof

Mansor

Baba

2015

2015 IEEE Student Conference on Research and Development (SCOReD)

View full text Add to dashboard Cite

show abstract

“…As Morimoto et al put it, "It would be difficult to naively apply the existing machine-learning methods to these robots" [19]. Thus, most of the work in this context has as the main focus the lowlevel kinematic control of the robot [20], [21], [22], [23], [24]. When dealing with low-level kinematics, data is not as sparse, and can often be further augmented by physical simulation (e.g., "envisioning").…”

Section: Introductionmentioning

confidence: 99%

Learning how to increase the chance of human-robot engagement

Macharet

Florêncio

2013

2013 IEEE/RSJ International Conference on Intelligent Robots and Systems

View full text Add to dashboard Cite

The increasing use of mobile robots in social contexts makes it important to provide them with the ability to behave in the most socially acceptable way possible. In this paper we investigate the problem of making a robot learn how to approach a person in order to increase the chance of a successful engagement. We propose the use of Gaussian Process Regression (GPR), combined with ideas from reinforcement learning to make sure the space is properly and continuously explored. In the proposed example scenario, this is used by the robot to predict the best decisions in relation to its position in the environment and approach distance, each one accordingly to a certain time of the day. Numerical simulations show a significant performance improvement when compared with a random technique. The robot is able to improve performance after just one day of interaction (a few dozens of trials), and achieves the maximum expected value for the proposed approach within sixty days.

show abstract