-
SURENA IV: Towards A Cost-effective Full-size Humanoid Robot for Real-world Scenarios
Authors:
Aghil Yousefi-Koma,
Behnam Maleki,
Hessam Maleki,
Amin Amani,
Mohammad Ali Bazrafshani,
Hossein Keshavarz,
Ala Iranmanesh,
Alireza Yazdanpanah,
Hamidreza Alai,
Sahel Salehi,
Mahyar Ashkvari,
Milad Mousavi,
Milad Shafiee Ashtiani
Abstract:
This paper describes the hardware, software framework, and experimental testing of SURENA IV humanoid robotics platform. SURENA IV has 43 degrees of freedom (DoFs), including seven DoFs for each arm, six DoFs for each hand, and six DoFs for each leg, with a height of 170 cm and a mass of 68 kg and morphological and mass properties similar to an average adult human. SURENA IV aims to realize a cost…
▽ More
This paper describes the hardware, software framework, and experimental testing of SURENA IV humanoid robotics platform. SURENA IV has 43 degrees of freedom (DoFs), including seven DoFs for each arm, six DoFs for each hand, and six DoFs for each leg, with a height of 170 cm and a mass of 68 kg and morphological and mass properties similar to an average adult human. SURENA IV aims to realize a cost-effective and anthropomorphic humanoid robot for real-world scenarios. In this way, we demonstrate a locomotion framework based on a novel and inexpensive predictive foot sensor that enables walking with 7cm foot position error because of accumulative error of links and connections' deflection(that has been manufactured by the tools which are available in the Universities). Thanks to this sensor, the robot can walk on unknown obstacles without any force feedback, by online adaptation of foot height and orientation. Moreover, the arm and hand of the robot have been designed to grasp the objects with different stiffness and geometries that enable the robot to do drilling, visual servoing of a moving object, and writing his name on the white-board.
△ Less
Submitted 30 August, 2021;
originally announced August 2021.
-
ODE-based Deep Network for MRI Reconstruction
Authors:
Ali Pour Yazdanpanah,
Onur Afacan,
Simon K. Warfield
Abstract:
Fast data acquisition in Magnetic Resonance Imaging (MRI) is vastly in demand and scan time directly depends on the number of acquired k-space samples. The data-driven methods based on deep neural networks have resulted in promising improvements, compared to the conventional methods, in image reconstruction algorithms. The connection between deep neural network and Ordinary Differential Equation (…
▽ More
Fast data acquisition in Magnetic Resonance Imaging (MRI) is vastly in demand and scan time directly depends on the number of acquired k-space samples. The data-driven methods based on deep neural networks have resulted in promising improvements, compared to the conventional methods, in image reconstruction algorithms. The connection between deep neural network and Ordinary Differential Equation (ODE) has been observed and studied recently. The studies show that different residual networks can be interpreted as Euler discretization of an ODE. In this paper, we propose an ODE-based deep network for MRI reconstruction to enable the rapid acquisition of MR images with improved image quality. Our results with undersampled data demonstrate that our method can deliver higher quality images in comparison to the reconstruction methods based on the standard UNet network and Residual network.
△ Less
Submitted 27 December, 2019;
originally announced December 2019.
-
Deep Plug-and-Play Prior for Parallel MRI Reconstruction
Authors:
Ali Pour Yazdanpanah,
Onur Afacan,
Simon K. Warfield
Abstract:
Fast data acquisition in Magnetic Resonance Imaging (MRI) is vastly in demand and scan time directly depends on the number of acquired k-space samples. Conventional MRI reconstruction methods for fast MRI acquisition mostly relied on different regularizers which represent analytical models of sparsity. However, recent data-driven methods based on deep learning has resulted in promising improvement…
▽ More
Fast data acquisition in Magnetic Resonance Imaging (MRI) is vastly in demand and scan time directly depends on the number of acquired k-space samples. Conventional MRI reconstruction methods for fast MRI acquisition mostly relied on different regularizers which represent analytical models of sparsity. However, recent data-driven methods based on deep learning has resulted in promising improvements in image reconstruction algorithms. In this paper, we propose a deep plug-and-play prior framework for parallel MRI reconstruction problems which utilize a deep neural network (DNN) as an advanced denoiser within an iterative method. This, in turn, enables rapid acquisition of MR images with improved image quality. The proposed method was compared with the reconstructions using the clinical gold standard GRAPPA method. Our results with undersampled data demonstrate that our method can deliver considerably higher quality images at high acceleration factors in comparison to clinical gold standard method for MRI reconstructions. Our proposed reconstruction enables an increase in acceleration factor, and a reduction in acquisition time while maintaining high image quality.
△ Less
Submitted 6 September, 2019; v1 submitted 30 August, 2019;
originally announced September 2019.
-
Non-Learning based Deep Parallel MRI Reconstruction (NLDpMRI)
Authors:
Ali Pour Yazdanpanah,
Onur Afacan,
Simon K. Warfield
Abstract:
Fast data acquisition in Magnetic Resonance Imaging (MRI) is vastly in demand and scan time directly depends on the number of acquired k-space samples. Recently, the deep learning-based MRI reconstruction techniques were suggested to accelerate MR image acquisition. The most common issues in any deep learning-based MRI reconstruction approaches are generalizability and transferability. For differe…
▽ More
Fast data acquisition in Magnetic Resonance Imaging (MRI) is vastly in demand and scan time directly depends on the number of acquired k-space samples. Recently, the deep learning-based MRI reconstruction techniques were suggested to accelerate MR image acquisition. The most common issues in any deep learning-based MRI reconstruction approaches are generalizability and transferability. For different MRI scanner configurations using these approaches, the network must be trained from scratch every time with new training dataset, acquired under new configurations, to be able to provide good reconstruction performance. Here, we propose a new generalized parallel imaging method based on deep neural networks called NLDpMRI to reduce any structured aliasing ambiguities related to the different k-space undersampling patterns for accelerated data acquisition. Two loss functions including non-regularized and regularized are proposed for parallel MRI reconstruction using deep network optimization and we reconstruct MR images by optimizing the proposed loss functions over the network parameters. Unlike any deep learning-based MRI reconstruction approaches, our method doesn't include any training step that the network learns from a large number of training samples and it only needs the single undersampled multi-coil k-space data for reconstruction. Also, the proposed method can handle k-space data with different undersampling patterns, and the different number of coils. Experimental results show that the proposed method outperforms the current state-of-the-art GRAPPA method and the deep learning-based variational network method.
△ Less
Submitted 18 March, 2019; v1 submitted 6 August, 2018;
originally announced August 2018.
