2,755 Hits in 6.6 sec

Deep Multi-Layer Perception Based Terrain Classification for Planetary Exploration Rovers

Chengchao Bai, Jifeng Guo, Linli Guo, Junlin Song
2019 Sensors  
Secondly, a deep neural network based on multi-layer perception is designed to realize classification of different terrains.  ...  Firstly, the time-frequency domain transformation of vibration information is realized by fast Fourier transform (FFT), and the characteristic representation of vibration information is given.  ...  Multi-Layer Perception Deep Neural Network Design Deep Neural Network Feedforward neural networks are a special form of supervised neural networks that use highprecision approximate computational models  ... 
doi:10.3390/s19143102 fatcat:vu6o3lbotfbgfnpfazwsveliky

Adaptive and intelligent navigation of autonomous planetary rovers — A survey

Cuebong Wong, Erfu Yang, Xiu-Tian Yan, Dongbing Gu
2017 2017 NASA/ESA Conference on Adaptive Hardware and Systems (AHS)  
2017) Adaptive and intelligent navigation of autonomous planetary rovers -a survey. In: 2017 NASA/ESA Conference on Adaptive Hardware and Systems, AHS 2017. IEEE, Piscataway, N.J., pp. 237-244.  ...  Reducing the reliance on human experts for navigational tasks on Mars remains a major challenge due to the harsh and complex nature of the Martian terrains.  ...  Using visual images of Martian terrain captured both on-ground and in-orbit, a deep convolutional neural network (CNN) was employed to classify terrain types and features.  ... 
doi:10.1109/ahs.2017.8046384 dblp:conf/ahs/WongYYG17 fatcat:iwlth237anf7dhrbpy3seuul3q

Phase-functioned neural networks for character control

Daniel Holden, Taku Komura, Jun Saito
2017 ACM Transactions on Graphics  
We present a real-time character control mechanism using a novel neural network architecture called a Phase-Functioned Neural Network.  ...  Recent developments in deep learning and neural networks have shown some promise in potentially resolving these issues.  ...  Deep Learning has found that it is possible to use input parameterizations which are not hand crafted in this way but use more neural network layers to perform the abstraction.  ... 
doi:10.1145/3072959.3073663 fatcat:jsmhirwlwfc2lfrjwx5vd2unba

Reinforcement Learning with Evolutionary Trajectory Generator: A General Approach for Quadrupedal Locomotion [article]

Haojie Shi, Bo Zhou, Hongsheng Zeng, Fan Wang, Yueqiang Dong, Jiangyong Li, Kang Wang, Hao Tian, Max Q.-H. Meng
2021 arXiv   pre-print
Unlike prior methods that use a fixed trajectory generator, the generator continually optimizes the shape of the output trajectory for the given task, providing diversified motion priors to guide the policy  ...  We then optimize the trajectory generator and policy network alternatively to stabilize the training and share the exploratory data to improve sample efficiency.  ...  [12] introduces an adaptation module to facilitate fast adaption on uneven terrains, by predicting the environmental factors based on the history of states and actions.  ... 
arXiv:2109.06409v2 fatcat:34g7jzc5cjcq3fw6seomdhqnn4

RADAR 2019 Author Index

2019 2019 International Radar Conference (RADAR)  
Continuous Wave Radar Using Deep Convolutional Neural Networks submission_65 SANTRA Avik Towards Adaptive MIMO Radar -ReceiverProcessing for Orthogonally Coded FMCWWaveforms submission_68  ...  Frequency Modulated Continuous Wave Radar Using Deep Convolutional Neural Networks submission_65 HIMED Braham Target Localization in Multi-static Passive Radar Systems with Artificial Neural Networks  ... 
doi:10.1109/radar41533.2019.9078992 fatcat:qgj7mi5yrfc7ti5qz6he5n4xvm

A deep learning framework for character motion synthesis and editing

Daniel Holden, Jun Saito, Taku Komura
2016 ACM Transactions on Graphics  
To map from high level parameters to the motion manifold, we stack a deep feedforward neural network on top of the trained autoencoder.  ...  Once motion is generated it can be edited by performing optimization in the space of the motion manifold.  ...  Locomotion on the Terrain The feedforward network is trained such that a curve drawn on the terrain is used to generate the actual locomotion of the character.  ... 
doi:10.1145/2897824.2925975 fatcat:tlw7vclqknawzdzsnqimtt2apy

Terrain-adaptive locomotion skills using deep reinforcement learning

Xue Bin Peng, Glen Berseth, Michiel van de Panne
2016 ACM Transactions on Graphics  
Figure 1 : Terrain traversal using a learned actor-critic ensemble. The color-coding of the center-of-mass trajectory indicates the choice of actor used for each leap.  ...  Building on recent progress in deep reinforcement learning (DeepRL), we introduce a mixture of actor-critic experts (MACE) approach that learns terrainadaptive dynamic locomotion skills using high-dimensional  ...  Deep Neural Networks (DNNs): Control policies based on DNNs have been learned to control motions such as swimming [Grzeszczuk et al. 1998 ], as aided by a differentiable neural network approximation of  ... 
doi:10.1145/2897824.2925881 fatcat:b2n5ytpbqzczll2lj5adz7tjjm

Insect-Inspired Robots: Bridging Biological and Artificial Systems

Poramate Manoonpong, Luca Patanè, Xiaofeng Xiong, Ilya Brodoline, Julien Dupeyroux, Stéphane Viollet, Paolo Arena, Julien R. Serres
2021 Sensors  
These interconnected and interdependent areas are all crucial to improving the level of performance of hexapod robotics in terms of energy efficiency, terrain adaptability, autonomy, and operational range  ...  We will also discuss how the next generation of bioroboticists will be able to transfer knowledge from biology to robotics and vice versa.  ...  We are grateful to the two anonymous editors, whose suggestions helped us improve the manuscript. Conflicts of Interest: The authors declare no conflict of interest.  ... 
doi:10.3390/s21227609 pmid:34833685 pmcid:PMC8623770 fatcat:7gne3leewjhwhaqs4bcd5hmxjm

Learning-Based End-to-End Path Planning for Lunar Rovers with Safety Constraints

Xiaoqiang Yu, Ping Wang, Zexu Zhang
2021 Sensors  
In addition, to improve the generalization ability to different lunar surface topography and different scale environments, a variety of training scenarios were set up to train the network model using the  ...  Firstly, a training environment integrating real lunar surface terrain data was built using the Gazebo simulation environment and a lunar rover simulator was created in it to simulate the real lunar surface  ...  Figure 5 . 5 The deep neural network architecture. Figure 5 . 5 The deep neural network architecture.  ... 
doi:10.3390/s21030796 pmid:33504073 pmcid:PMC7866010 fatcat:cqxcxbnxq5arxa5mfkixwtdswe

GCCE 2020 Subject Index

2020 2020 IEEE 9th Global Conference on Consumer Electronics (GCCE)  
T U V W Self-Attention Based Neural Network for Few Shot Classification Self-Attention Based Neural Network for Few Shot Classification Separation of Multiple Sound Sources in the Same Direction  ...  by Instantaneous Spectral Estimation Separation of Multiple Sound Sources in the Same Direction by Instantaneous Spectral Estimation Sequence-To-One Neural Networks for Japanese Dialect Speech Classification  ...  Deep Learning Using Generative Adversarial Networks Data Augmentation for Deep Learning Using Generative Adversarial Networks Data Augmentation Using User Attention for Educational Content Recommendation  ... 
doi:10.1109/gcce50665.2020.9291796 fatcat:bmnnn7xnxrefhaneq262fe4i6u

Recent developments in terrain identification, classification, parameter estimation for the navigation of autonomous robots

M. G. Harinarayanan Nampoothiri, B Vinayakumar, Youhan Sunny, Rahul Antony
2021 SN Applied Sciences  
The study focuses on various Deep Learning techniques and Fuzzy Logic Systems in detail. The work can be extended to develop new control schemes to improve multiple terrain navigation performance.  ...  AbstractThe work presents a review on ongoing researches in terrain-related challenges influencing the navigation of Autonomous Robots, specifically Unmanned Ground ones.  ...  Sensor data is processed using FFT (Fast Fourier Transform). A probabilistic neural network is used to obtain the terrain information based on feature extraction by FFT. Giguire et al.  ... 
doi:10.1007/s42452-021-04453-3 fatcat:zftqt26h7fgsblkmwduo336u4y

Real-time Optimal Navigation Planning Using Learned Motion Costs

Bowen Yang, Lorenz Wellhausen, Takahiro Miki, Ming Liu, Marco Hutter
2021 2021 IEEE International Conference on Robotics and Automation (ICRA)  
A GPU-aided, sampling-based path planner combined with a gradient-based path optimizer provides optimal paths by using a neural network-based locomotion cost predictor which is trained in simulation.  ...  Navigation on challenging terrain topographies requires the understanding of robots' locomotion capabilities to produce optimal solutions.  ...  Nevertheless, it requires a long planning time to successively predict a large number of sampled motions for a smooth and feasible path using a deep neural network.  ... 
doi:10.1109/icra48506.2021.9561861 fatcat:bevyromdcvhsjn6itl664wjerq

Neural Network Model for Path-Planning of Robotic Rover Systems [article]

Youssef Bassil
2012 arXiv   pre-print
The network is trained in offline mode using back-propagation supervised learning algorithm.  ...  Today, robotics is an auspicious and fast-growing branch of technology that involves the manufacturing, design, and maintenance of robot machines that can operate in an autonomous fashion and can be used  ...  [16] proposed an adaptive approach for controlling robot manipulators using neural networks.  ... 
arXiv:1204.0183v1 fatcat:cncikjpkijdutmd525xeu7ebja

Decentralized control and local information for robust and adaptive decentralized Deep Reinforcement Learning

Malte Schilling, Andrew Melnik, Frank Ohl, Helge Ritter, Barbara Hammer
2021 Neural Networks  
Finally, when examining generalization to uneven terrains-not used during training-we find best performance for an intermediate architecture that is decentralized, but integrates only local information  ...  This appears as a promising approach towards more robust DRL and better generalization towards adaptive behavior.  ...  In DRL, deep neural networks are used as a function approximator that maps a currently observed state to actions or action probabilities (Arulkumaran et al., 2017) .  ... 
doi:10.1016/j.neunet.2021.09.017 pmid:34673323 fatcat:24thoeeh3naubl5fk2rfxtk7ta

Mars Terrain Segmentation with Less Labels [article]

Edwin Goh, Jingdao Chen, Brian Wilson
2022 arXiv   pre-print
This research proposes a semi-supervised learning framework for Mars terrain segmentation where a deep segmentation network trained in an unsupervised manner on unlabeled images is transferred to the task  ...  The network incorporates a backbone module which is trained using a contrastive loss function and an output atrous convolution module which is trained using a pixel-wise cross-entropy loss function.  ...  On the other hand, methods such as network adaptation [28] and correlation alignment [29] enable deep neural networks to better generalize to the target domain based on introducing additional loss  ... 
arXiv:2202.00791v1 fatcat:zlmamlrj7festiiy2yq2z6ffay
« Previous Showing results 1 — 15 out of 2,755 results