The assembly of earphone parts in electro acoustic factory automation is a challenging problem for industrial robots. The main difficulties come from the fact that conventional industrial robots perform poorly on rapi...
The assembly of earphone parts in electro acoustic factory automation is a challenging problem for industrial robots. The main difficulties come from the fact that conventional industrial robots perform poorly on rapid-updated earphone parts, which are characterized by small size, variety and irregular shapes. What's worse, the robot cannot adapt to the earphone parts with different poses effectively. In order to tackle above problems, an adaptable robotic system guided by the 3D laser triangulation sensor and six-axis F/T sensor is built in this paper. In addition, to reduce position errors caused by the 3D sensor accidently, an assembly strategy which is implemented by the admittance controller learned from human teaching, is proposed for assembling earphone parts. Specifically, AdaBoost-based GPR method is applied to derive the admittance gain, so that the controller perform effectively. A series of comparative experiments prove that the established system could demonstrate superior performance on the earphone parts with small size, variety and irregular shapes. Furthermore, it is also validated that the assembly strategy could be executed effectively on the earphone parts with different poses.
Due to poor robustness, instability, and the heavy burden of use, the traditional myoelectric control method is still powerless in the face of the control of the dexterous prosthetic hand. To solve this problem, a new...
Due to poor robustness, instability, and the heavy burden of use, the traditional myoelectric control method is still powerless in the face of the control of the dexterous prosthetic hand. To solve this problem, a new method (CVEEI), that combines computer vision, eye tracking, electromyogram (EMG) and IMU was proposed in this paper. Firstly, through gazing (eye-tracking) in front of the screen, the grasping pattern of the objects can be fed back to the prosthetic hand controller; Then, the prosthetic hand can be controlled to transport the object to the position expected, on collaboration of both EMG and IMU. In this process, the grasping pattern of all objects can be recognized by computer vision in real-time. Importantly, through comparing the traditional EMG control method (co-contraction to switch) in the transport experiment of the objects, the superiority of this new method in operating the dexterous prosthetic hand was further verified (fast > 1s/single object) in this paper.
Two-wheeled robots have many advantages over other mobile robots, but they are difficult to self-balance compared with other wheeled robots. Reinforcement learning (RL) is a general framework for sequential decision-m...
Two-wheeled robots have many advantages over other mobile robots, but they are difficult to self-balance compared with other wheeled robots. Reinforcement learning (RL) is a general framework for sequential decision-making problems. So far, there are many applications of reinforcement learning to solve robot control problems, but most of them are used in simulators because of the large amount of data required. In addition, due to the reality gap, the policy learned in a simulation environment cannot be transferred directly to a real robot. Real robot data often expensive due to the potential damage to the robot. Model-based methods require far fewer robot data than model-free methods, but these methods have the problem of model bias. In this paper, we use a model-based reinforcement learning method to achieve self-balance of a two-wheeled robot. We present a model learning method that can reduce the problem of model bias. Our method combines the simulator and a few real robot data to learn a probabilistic dynamics model of the robot through an iterative way, which requires no expertise and can learn from scratch. Then the control policy is optimized based on the learned model.
The minimally invasive surgical robot greatly improved surgical OR efficiency, and surgeon manipulates the surgical instruments through master-slave control under the guidance of laparoscope vision. As a mapping of th...
The minimally invasive surgical robot greatly improved surgical OR efficiency, and surgeon manipulates the surgical instruments through master-slave control under the guidance of laparoscope vision. As a mapping of the surgeon's hand-eye coordination, the coordination between surgical instruments and laparoscope has an important impact on the surgical efficiency and the ergonomics. In this paper, a novel hand-eye coordination algorithm is proposed for the robot -assisted minimally invasive surgery (RMIS). Firstly, a novel cascade-calibration (CC) algorithm structure is proposed to determine the pose relationship between the robot laparoscope arm coordinate system and the robot instruments arm coordinate system. A hand-eye coordination algorithm is also proposed, which provides accurate master-slave control and improves the surgical operation room (OR) efficiency. We validate the proposed algorithm through two experiments with our minimally invasive surgical robot system. Our experiments demonstrate that the proposed method is competitively effective to induce the hand-eye coordination for the RMIS.
Accurate placement of the needle is critical in percutaneous surgery. Needle shape reconstruction technology based on fiber Bragg gratings (FBGs) sensor is considered to have the potential to achieve this goal. In thi...
Accurate placement of the needle is critical in percutaneous surgery. Needle shape reconstruction technology based on fiber Bragg gratings (FBGs) sensor is considered to have the potential to achieve this goal. In this paper, we present a temperature-insensitive calibration model for calibrating needles with FBG sensors to accurately estimate its deformation. In addition, we have designed a device for simultaneously calibrating the loading direction and shape. Preliminary results indicate that the proposed model is theoretically insensitive to temperature, but in fact it can only be used within a certain temperature range. The resolution of the loading direction increases as the degree of bending increases, with a maximum error of 8.58°. Shape reconstruction error can be less than 1.5mm with small needle bending. This calibration method can meet clinical applications.
As the robot gradually develops from indoor to outdoor, multi-sensor fusion is increasingly being applied to the field of robot perception. Among them, LiDAR and binocular sensors are most widely used. Robots use the ...
As the robot gradually develops from indoor to outdoor, multi-sensor fusion is increasingly being applied to the field of robot perception. Among them, LiDAR and binocular sensors are most widely used. Robots use the complementary and redundant data they provide to perform environmental perception and implement certain functions of the robot, such as SLAM, autonomous obstacle avoidance and so on. However, the data obtained by each sensor are relative to its own coordinate system. Data must be converted to a unified coordinate system before data fusion. Therefore, the calibration of the transformation among sensors is the first problem to be solved in data fusion. Moreover, the accuracy of calibration is directly related to the effect of data fusion. Therefore the calibration between sensors is very important and necessary. In this paper, we propose a novel method to find accurate rigid-body transformation for the extrinsic calibration of a 3D LiDAR and a stereo cameras, using two Aruco calibration boards. Also the validity of the algorithm is demonstrated by some experiments.
It is a question of great import for the dexterous prosthesis to realize its continuous, simultaneous and proportional control of multi-DOFs so that its rehabilitation function can be well achieved and its user accept...
It is a question of great import for the dexterous prosthesis to realize its continuous, simultaneous and proportional control of multi-DOFs so that its rehabilitation function can be well achieved and its user acceptance rate can be improved. In this paper, we propose a new method, CNMF-HP, which integrates the constrained non-negative matrix factorization (NMF) and the Hadamard product, for estimating 2-DOF wrist movements (DOF-1, flexion/extension; DOF-2, adduction/abduction) from myoelectric signals. Based on this method, the simultaneous control of a multi-fingered prosthesis is verified in this paper. Through establishing the mapping relationship between the wrist movements and the finger movements of the dexterous prosthesis, two grasp control strategies (synchronous and asynchronous) are proposed and successfully applied in operating the dexterous prosthesis to grasp four kinds of objects of different grasping patterns (cylindrical, spherical, tripod and lateral). Then, the performance of different strategies (synchronous/asynchronous control) in the grasping process is compared in terms of the completion time and completion rate of the tasks, in this way the superiority of EMG control proposed in this paper is further verified.
Aiming at the shortcomings of dynamic windowing algorithm (DWA) into local optimal solution. This work studys a modified DWA algorithm integrated global path planning. By introducing the result of global path planning...
Aiming at the shortcomings of dynamic windowing algorithm (DWA) into local optimal solution. This work studys a modified DWA algorithm integrated global path planning. By introducing the result of global path planning as a reference trajectory, a novel evaluation function is designed which makes the optimal trajectory be guaranteed. In addition, the work also proposes the evaluation sub-function of direction, the evaluation sub-function of smoothing speed and the evaluation sub-function of acceleration, separately, ensures the direction, smoothness and speed of movement. The algorithm is verified in mobile robot, experimental results show that the algorithm optimizes the path of mobile robot, simultaneously, the advantages of the traditional DWA, such as the practicability and continuity of motion are retained.
In this paper, the servo system mechanical drive is analyzed, flexible beam vibration will cause end point trajectory distortion. Simplify the flexible beam load as a second-order system, which is a low frequency ligh...
详细信息
ISBN:
(纸本)9781728133997
In this paper, the servo system mechanical drive is analyzed, flexible beam vibration will cause end point trajectory distortion. Simplify the flexible beam load as a second-order system, which is a low frequency light damped vibration system, the ZVD input shaper which is very suitable for vibration suppression of low frequency light damping system is adopted. Due to the different vibration frequencies and damping coefficients of the X-Y platform flexible load, the X-axis and Y-axis input shapers cause different command lags, which can cause trajectory distortion. The adaptive feedforward controller based on phase angle error is used to solve trajectory distortion problem. Combined with the input shaper and adaptive feedforward control based on phase error, it can solve the vibration suppression of X-axis and Y-axis flexible load with different vibration characteristics, and can guarantee the track accuracy. Finally, the effectiveness of the algorithms is verified by simulation.
This paper presents a highly compliant prosthetic hand based on a novel synergy mechanism. The prosthetic hand consists of a synergy mechanism, four fingers and one thumb. The synergy mechanism can transmit the power ...
This paper presents a highly compliant prosthetic hand based on a novel synergy mechanism. The prosthetic hand consists of a synergy mechanism, four fingers and one thumb. The synergy mechanism can transmit the power from two motors to four fingers. It can effectively reduce the number of motors. At the same time, it makes the prosthetic hand have excellent dexterity. The underactuated finger has three joints with two DOF. It has two main motion modes: coupled motion in free space and self-adaptive motion when contacting with the object, which can mimic the human finger as much as possible. The thumb uses one motor to drive the two flexion-extension joints and uses a manual switch to drive the abduction-adduction joint, which can reduce the number of motors and the cost efficiently. The performance evaluation is given to demonstrate the comprehensive performance in terms of the versatility, compliance, sensing, size, weight and cost of the proposed design.
暂无评论