Ravi Balasubramanian

Reinforcement learning (RL) has demonstrated success across multiple robotic grasping and manipulation tasks. However, for RL to be widely applicable, policies must be able to transfer across the sim-to-real gap, and transfer to hand geometries that they are not trained on. Methods such as domain randomization and domain adaptation only partially help with bridging these gaps. In this letter, we explore the impact of state and action space selection on transferability across both the sim-to-real gap and across different hand geometries. Using two exemplar manipulation tasks we demonstrate that state and action space selection significantly affect the overall performance of a policy and its robustness to both types of transfer. We also show that, for both types of transfer, a reduced state space that avoids hand specific information is preferable, even when it provides less information than a full state space.

Tendon transfer surgery is often used to restore hand grasp function following high median-ulnar nerve palsy. This surgery typically reroutes and sutures the tendon of the extensor carpi radialis longus (ECRL) muscle to all four flexor digitorum profundus (FDP) tendons of the hand, coupling them together. This makes it difficult to grasp irregularly shaped objects. We propose inserting a novel implantable passive device between the FDP tendons to surgically construct a differential mechanism, enabling the fingers to individually adapt to the irregular contours during grasping. These passive implants with no moving parts are fabricated from biocompatible materials. We tested the implants’ ability to create differential flexion between the index and middle fingers when actuated by a single muscle in two human cadaver hands using a computerized closed-loop control paradigm. In these cadaveric models, the implants enabled significantly more differential flexion between the index and middle fingers for a wide range of donor tendon tensions. The implants also redistributed fingertip forces between fingers. When grasping uneven objects, the difference in contact forces between fingers reduced by nearly 23% compared to the current suture-based surgery. These results suggest that self-adaptive grasp is possible in tendon transfers that drive multiple distal flexor tendons.

Deep Learning is a subdivision of machine learning and Artificial intelligence (AI). Autonomous Deep learning enables human brain to think and learn computers. In recent days Deep learning is used in many domains, especially in medical field. It is used mainly in classification. The Convolutional Neural Network (CNN) is one among the best technique in DL. It is best suitable in image classifications. CNN is directed to process the data into multiple layers of arrays. It is used for computationally efficient. Brain Tumor is one of the dangerous diseases in India as well as the whole world. A brain tumor is an unwanted cell in the brain. Brain tumor symptoms are based on size, location and type .There are two types of brain tumor. Brain tumor tissue affects on the brain that is called primary tumor. Brain tumor tissue affects in outside the brain that is called as secondary tumor (metastatic).In this paper, we are analyzing various Deep Convolution Neural Network on brain tumor perspectives. Here, LeNet, AlexNet, ResNet-18, VGG Net-16 are discussed and Evaluation metrics like Accuracy, F1 score, Precision, Recall are used to identify the performance of the above techniques.

This letter presents a novel benchmark called the Asterisk Test that characterizes a hand’s ability to translate an object on a table in eight linear directions using only two fingers. The benchmark is agnostic to the controller, object, and robot hand used. The benchmark also enables a measure of the symmetry (about the plane’s principal axes) of the hand’s ability to maneuver objects. The Asterisk Test’s utility is demonstrated on two common robot hand designs: a parallel jaw gripper and a two-linked, two fingered gripper. The method to analyze the performance data in order to evaluate the hand is also provided. We conclude with a comparison of this benchmark to previous dexterity measures, such as manipulability. We also discuss how improved performance in this benchmark is indicative of simpler hand control.

Robotics has the potential to enable custom, small-scale manufacturing, and the ability to automate many tasks (such as part clean up) that are currently performed manually. As part of this effort, we need the ability to quickly and efficiently evaluate the quality of that clean up, preferably without requiring expensive laser scanners. In this paper, we focus on visual evaluation

Robotics has the potential to enable custom, small-scale manufacturing, and the ability to automate many tasks (such as part clean up) that are currently performed manually. As part of this effort, we need the ability to quickly and efficiently evaluate the quality of that clean up, preferably without requiring expensive laser scanners. In this paper, we focus on visual evaluation

Objective: A tendon-transfer is a reconstructive orthopedic surgery where tendons are re-routed from a non-functioning muscle and attached a functioning muscle. Prior work has shown that using a passive implanted device in the ECRL-to-FDP tendon-transfer surgery significantly improves hand grasping function. However, it is still unclear how hand-function improvement, measured by finger joint range of motion and torque, is dependent on the implant’s geometry and location within the tendon network. This paper presents a genetic algorithm that determines the device’s optimal geometry and location. Methods: Hand biomechanical simulation platform was developed to model hand function and also model the tendon-transfer surgery with and without the implant. Finger kinematics and joint torque were used to develop three unique objective functions to optimize the implant’s parameters. Results: The optimized device resulted in an 11X increase in finger kinematics with only a 0.9% decrease in joint torque when compared with the biomechanical function enabled by the current suture-based surgery. Conclusion: Designing implantable devices that modify musculoskeletal function is challenging. Factors like tendon routing and joint kinematics create a complex nonlinear system when considering biomechanical function. A genetic algorithm is an effective tool to tackle these nonlinear landscapes to produce optimized designs. Significance: The state-of-the-art surgical procedure to repair high median-ulnar nerve palsy leads to poor hand function and severely limits the patient’s ability to perform activities of daily life. This work provides a method for defining relevant objective functions for hand biomechanical function and then uses those objective functions with genetic algorithms to optimize the geometry of an orthopedic implant across multiple variables. The achieved biomechanical function is significantly better than hand function enabled by current surgical procedure.

Image dehazing is an essential preprocessing stage for several applications such as surveillance, long-range imaging, automatic driver-assistance system (ADAS), and remote sensing. Haze particles degrade visible-band images more severely than the infrared images due to scattering phenomena. Therefore, infrared data can be effectively utilized to enhance the visibility of color images captured in a hazy environment. In this brief, for the first time, a multispectral transmission map fusion approach is presented, wherein a haze-aware weight generation scheme is used to improve transmission map in specific regions. This achieves significant quantitative improvement as compared with the existing methods. In addition, the method is extremely suitable for hardware implementation. Therefore, a low-cost very large-scale integration (VLSI) architecture is also presented. The application-specified integrated circuit (ASIC) and field-programmable gate array (FPGA) implementations dissipate 3.45- and 57-mW power at 150 MHz, respectively. Both ASIC and FPGA achieve a high throughput of 450 and 308 Mpixels/s, respectively. The design does not require any external memory such as dynamic random access memory, thus making it suitable for on-chip processing that can be closely integrated with an image sensor.

INTRODUCTION: In the United States, approximately 17,700 new cases of spinal cord injury occur each year, with 58.7% resulting in some degree of |