Assistive robots promise to be of great help to wheelchair users with motor impairments, for example for activities of daily living. Using shared control to provide task-specific assistance – for instance with the Sh...
详细信息
ISBN:
(数字)9798350384574
ISBN:
(纸本)9798350384581
Assistive robots promise to be of great help to wheelchair users with motor impairments, for example for activities of daily living. Using shared control to provide task-specific assistance – for instance with the Shared Control Templates (SCT) framework – facilitates user control, even with low-dimensional input signals. However, designing SCTs is a laborious task requiring robotic expertise. To facilitate their design, we propose a method to learn one of their core components – active constraints – from demonstrated end-effector trajectories. We use a probabilistic model, Kernelized Movement Primitives, which additionally allows adaptation from user commands to improve the shared control skills, during both design and execution. We demonstrate that the SCTs so acquired can be successfully used to pick up an object, as well as adjusted for new environmental constraints, with our assistive robot EDAN.
This work is devoted to the development of a software module for analyzing the textures of machined parts on CNC machines using computer vision methods. The main goal of the research is to create a tool capable of ana...
详细信息
On the basis of a non-classical approach to the solution of boundary value problems of statics, a mathematical model of deformation of elastic structures and their elements has been developed. This model is implemente...
详细信息
This paper presents two robust optimal controllers for Cooperative Manipulators (CoMans) holding an object on the basis of a modified factorization approach. Due to advantages such as: simplicity, flexibility, systema...
详细信息
In-hand manipulation with multi-fingered hands is a challenging problem that recently became feasible with the advent of deep reinforcement learning methods. While most contributions to the task brought improvements i...
详细信息
ISBN:
(数字)9798350373578
ISBN:
(纸本)9798350373585
In-hand manipulation with multi-fingered hands is a challenging problem that recently became feasible with the advent of deep reinforcement learning methods. While most contributions to the task brought improvements in robustness and generalization, this paper addresses the critical performance measure of the speed at which an in-hand manipulation can be performed. We present reinforcement learning policies that can perform in-hand reorientation significantly faster than previous approaches for the complex setting of goal-conditioned reorientation in $\mathrm{SO}(3)$ with permanent force closure and tactile feedback only (i.e., using the hand’s torque and position sensors). Moreover, we show how policies can be trained to be speed-adjustable, allowing for setting the average orientation speed of the manipulated object during deployment. To this end, we present suitable and minimalistic reinforcement learning objectives for time-optimal and speed-adjustable in-hand manipulation, as well as an analysis based on extensive experiments in simulation. We also demonstrate the zero-shot transfer of the learned policies to the real DLR-Hand II with a wide range of target speeds and the fastest dextrous in-hand manipulation without visual inputs. Website: https://***/manipulation/humanoids24
Recent improvements in the Artificial Intelligence and Machine Learning lead us to the automated models for detecting diseases based on ECGs. The previous models based on 2D CNNs had pretty good accuracy in ECG Classi...
Recent improvements in the Artificial Intelligence and Machine Learning lead us to the automated models for detecting diseases based on ECGs. The previous models based on 2D CNNs had pretty good accuracy in ECG Classification, but these models requires us to create a 2D spatial representation from ECGs as ECGs are 1D temporal data. As these are typically done through CWT Scalogram it is computationally intensive. In this paper a model based on a 1D Convolutional Neural Network is proposed which can classify specific cardiac disease from ECG data. This model consists of four feature extracting Convolutional layers each succeeded by a Max Pooling layer and a dropout layer to prevent overfitting. This model is trained with ECGs of different types based on different cardiac disease like Cardiac Arrhythmia, Congestive Heart Failure and Normal Sinus Rhythm etc. and it can classify these diseases from ECGs. For training and evaluation MIT-BIH arrhythmia database have been used. It results in classification accuracy of 96.83% which is almost equal to the previous 2D CNN model with significant decrease in complexity.
This paper presents the design of a 3D printed mobile differential drive robot with LiDAR and 3D depth sensing. The robot is designed to be low cost and easy to build, making it suitable for educational and research p...
详细信息
ISBN:
(数字)9798350372694
ISBN:
(纸本)9798350372700
This paper presents the design of a 3D printed mobile differential drive robot with LiDAR and 3D depth sensing. The robot is designed to be low cost and easy to build, making it suitable for educational and research purposes. All the main structures of the robot are 3D printed while the parts used in the robot are from off-the-shelf components. All the design files and source code used in this project is shared in the github repository. The robot is equipped with a LiDAR sensor for mapping and navigation, and a 3D depth sensing camera for object recognition. Robot Operating System (ROS) running on an Intel single board computer serves as the middleware for all robot control operations. The developed robot is able to run simultaneous localization and mapping (SLAM) and perform autonomous navigation. Design files and the software for the robot can be obtained from the github reposi-tory https://***/dannyngweekiat/3D-Printed-Differential-Drive-Bot.
Cranes suffer significantly from parameter uncertainties, frictions, unknown dynamics, and disturbances. Besides, it may be necessary to design a state estimator for an effective feedback controller design. Furthermor...
详细信息
ISBN:
(数字)9781665464543
ISBN:
(纸本)9781665464550
Cranes suffer significantly from parameter uncertainties, frictions, unknown dynamics, and disturbances. Besides, it may be necessary to design a state estimator for an effective feedback controller design. Furthermore, the performance quality and precision of model-based controllers and state estimators rely heavily on accurate models, which may be challenging to formulate. In this study, to address the highlighted control problems and also achieve a precise level of bogie tracking error, the sliding mode control (SMC) with a novel prescribed performance function-based sliding surface is designed. Simulations have shown that the proposed control technique achieved pre-specified transient and steady-state quality metrics on the bogie tracking error where the error converged to a pre-defined small vestigial set, with a decay rate no less than the pre-assigned rate, exhibiting overshoots and undershoots less than some small-enough pre-assigned values.
An important aspect of robots' control highly depends on the type of interface which drives the interaction between humans and machines. In the present work, the authors propose a personal implementation of a cont...
An important aspect of robots' control highly depends on the type of interface which drives the interaction between humans and machines. In the present work, the authors propose a personal implementation of a control method for a mechanical structure using an exoskeleton interface. This work presents each stage of development, from concept design to implementation and testing. In the final part of the paper, the design and experiment results are shown, reflecting benefits and future improvement possibilities.
As people tend to integrate technology more and more in their lives, one of the most evident aspects which has a great impact on their learning methods are the E-learning platforms which provide interactive means to d...
详细信息
ISBN:
(数字)9798350364293
ISBN:
(纸本)9798350364309
As people tend to integrate technology more and more in their lives, one of the most evident aspects which has a great impact on their learning methods are the E-learning platforms which provide interactive means to develop new skills in problem-solving topics. The development of an E-learning platform that simplifies the way in which users engage with the information while simultaneously enhancing their level of comprehension is, without a question, the fastest and most effective strategy. Trend, in matter of learning techniques, has also brought new challenges and opportunities to enhance assimilation of information. Thus, during the process of learning, focus has to be actively maintained through a series of stimuli. Algorithm visualization blended with Artificial Intelligence is the solution that the paper intends to present as a reaction to the actual inclination.
暂无评论