Please use this identifier to cite or link to this item:
|Title:||Depth-based obstacle avoidance through deep reinforcement learning||Authors:||Wu, Keyu
Mahdi Abolfazli Esfahani
|Keywords:||Engineering::Electrical and electronic engineering||Issue Date:||2019||Source:||Wu, K., Mahdi Abolfazli Esfahani, Yuan, S. & Wang, H. (2019). Depth-based obstacle avoidance through deep reinforcement learning. ICMRE'19: Proceedings of the 5th International Conference on Mechatronics and Robotics Engineering, 102-106. doi:10.1145/3314493.3314495||Abstract:||Obstacle avoidance is an indispensable technique for mobile robots to maneuver safely without collision. In this paper, we propose an end-to-end deep neural network to derive control commands directly from the raw depth images using deep reinforcement learning. The convolutional neural networks are used to extract the feature representation from the input depth images and the fully connected neural networks subsequently map the features to Q-values for determination of the optimal action. To improve the performance of the network, we adopt a two-stage method so that noisy fully connected layers are employed at the beginning while conventional ones are utilized during the second stage of training. Compared to the existing method, our proposed model exhibits much better performance in avoiding obstacles and converges faster during training.||URI:||https://hdl.handle.net/10356/142309||DOI:||10.1145/3314493.3314495||Rights:||© 2019 Association for Computing Machinery. This paper was published in ICMRE'19: Proceedings of the 5th International Conference on Mechatronics and Robotics and is made available with permission of Association for Computing Machinery.||Fulltext Permission:||open||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Conference Papers|
Files in This Item:
|Depth-based obstacle avoidance through deep reinforcement learning.pdf||786.46 kB||Adobe PDF||View/Open|
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.