Chin-Yin Chen

Localization and navigation using QR code for mobile robot in indoor environment

Localization and navigation are fundamental issues to mobile robotics. An approach for localization and navigation for a mobile robot in indoor environment is proposed in this paper. In this approach, QR codes are used as landmarks to provide global pose references. Label and location information is stored in QR codes which are strategically placed in the operating environment. The mobile robot is equipped with an industrial camera pointing to the ceiling to read QR codes at high speed. The pose of the robot is estimated according to the positional relationship between QR codes and the robot. For the purpose of collision-free navigation, a laser range finder (LRF) is applied to build a map in unknown environment and detect obstacles. Dijkstra algorithm and Dynamic Window Approach (DWA) are applied in path planning based on a 2D grid map. Experimental results show that this approach has good feasibility and effectiveness.

An intuitive teaching system using sensor fusion and 3D matching

In this paper, a fast intuitive system for robot teaching based on sensor fusion and 3D shapes matching technology is presented. The off-line robot programming technique has been developed for a few years. However, many intuitive ways are set up complex system which always need relatively high initial investment in software and workers training. In this paper, we also present a system to program an UR robot in an intuitive way, which allow users, even non-expert workers, to program robot more simple and faster, whats more, this system which is much easier to implement is simply and inexpensive, making it more vulnerable to be widely used, especially in small and medium enterprises. In this system, a human-robot interaction system has been designed, a 3D registration algorithm is proposed for the matching between virtual 3D model and real work piece. Once the registration is completed, the system can convert any cloud point data from the virtual 3D model into the corresponding points on the real work piece. In addition, the great deal of information included in the 3D model makes the tool orientation teaching more easily. We aim to create an easy-to-use robot programming system which could be accessible to anyone without professional skills, and this system is required to be high efficiency and precision. Additionally, its safe because of users do not need to be physically closed to the machine working environment.

A Dual-Loop Dual-Frequency Torque Control Method for Flexible Robotic Joint

The impedance control of most flexible joints generally has a cascaded structure with both inner torque feedback and outer position feedback loops. The torque control accuracy and bandwidth have great effects on their impedance control performance. But in practical applications, the improvement of the torque control performance are limited by many factors, such as the high noise level of torque sensor and the model uncertainty and nonlinearity. This paper proposes a dual-loop torque control structure with a dual-frequency control in the inner loop and a disturbance observer in the outer loop. The dual-frequency control uses a frequency-separated controller design method that employs two controllers in low and high frequency, respectively. The high frequency controller with relatively conservative gains is designed to ensure the system stability. Thus, the low frequency controller can focus on improving the torque control accuracy, without experiencing the limitations of system stability and torque sensor noise. Moreover, the disturbance observer is introduced to compensate the model uncertainty and nonlinearity. Simulations are conducted to verify the effectiveness of the proposed dual-loop dual-frequency torque controller.

Design and Control of Two Degree of Freedom Powered Caster Wheels Based Omni-Directional Robot

Omni-directional robot has the ability of 0–360° motions are received much attention in recent years. They have locomotive advantages and are widely deployed in larger range of application fields especially in constrained narrow space. This paper introduce a novel omni-directional robot with four powered caster wheels, each caster wheel has two degree of freedom (DOF) and made by two outer rotor motors connected mechanically. The kinematics of the system are analyzed, the prototype has been developed. The developed omni-directional robot is able to realize the moving motions along x and y axis and rotate about z-axis. All of the software are implemented on Robot Operation Systems (ROS) and the velocity trajectory planner is employed and embedded in the software. The squared position curve is given and tested by using lase tracker, the result is analyzed, and it shown that the following error of the system is about 2.5 cm while the width of the square is 70 cm.

Using 3D Matching for Picking and Placing on UR Robot

Due to conventional industrial robot moving has been known that programing is either tedious or simplex and operating platform requires a pure environment without unnecessary distractions from other objects. This paper presents a novel method which using shape based 3Dmatching technology for picking and placing object by UR robot. This method is able to help the robot catch a specified object in any complex environment. To do this, we need only one camera and the targets 3D CAD model. There are many efforts was indeed carried out in order to improve the control accuracy, first , the 3D coordinates of the interested points should be high-resolution calculated with the images that areprovided by the camera and we use a special method to optimize internal and external camera parameters when calibrating the camera, Then, A 3D-matching-operator is used to search for the specific objects in the real-time images based on the given 3D model, which is realized by projects the edges of the 3D object model that was used to create the 3D shape model into the image coordinate system and return projected edges. Finally, the coordinates of the 3D object model is obtained in the PTU coordinate system through the integration of a PTU and a Laser range finder , and a transformational matrix is obtained for calculate the coordinates of the targets on UR robot’s base coordinate system. Based on this technology, users only simply need to import a 3D CAD model and click on the image of the workplace to define the end point, the robot will exceed the procedure of pick and place automatically.

Cartesian admittance control with on-line gravity and friction observer compensation for elastic joint robots

In this paper, a Cartesian admittance controller with on-line gravity and friction observer compensation based on passivity theory for elastic joint robots is proposed. For this study, a compliance and position controller for joint and Cartesian level has been described with dual loop construction: an outer-loop admittance control, and an inner-loop motor side position with torque feedback control. In terms of torque feedback, a physical interpretation is addressed and analyzed. Regarding as admittance controller of the elastic joint, two type (stiffness and damping) of the systematic control model can be represented. In addition to an on-line gravity compensation based-on output-side encodes has been proposed. Moreover, in order to achieve high trajectory tracking performance, a friction observer based compensation is applied also. Furthermore, aimed at the control performance of the stiffness, damping, and admittance control algorithm are assessed by a simulation, respectively. Finally, the simulation studies reveal a better effect regarding as the position tracking and torque output in contrast to without friction passivity algorithm and proportional-derivative (PD) control. In addition, the response of proposed compliance control method is effective for external force.

A geometrical work piece localization algorithm for rapid robot programming

When an offline programming method is employed for an industrial robot, accurate work piece localization is critical as there always exist discrepancies between the actual and nominal work piece locations. By drawing upon the mathematical tools from group theory and differential geometry, three fundamental error models (i.e., point-to-point, line-to-line, and plane-to-plane models) are formulated to describe the geometrical “distances” between two 3D models of the same work piece, which are obtained from the original CAD design and the in-situ measurement data, respectively. Consequently, the work piece localization becomes a 3D registration problem, i.e., to determine the optimal kinematic transformation between the two 3D models through minimizing their geometrical “distances”. Based on the proposed error models, a least-square algorithm is employed to iteratively identify the kinematic transformation parameters. Simulation results are provided to illustrate the effectiveness of the proposed algorithms.

An improved FastSLAM algorithm based on an omni-directional wheeled mobile robot

The indoor mobile robots have been widely employed in manufacturing and service industries in recently years. However, the conventional simultaneous localization and mapping (SLAM) method is mainly meant for mobile robots with traditional drives (differential drive and Ackerman steering) that have the coupled translation and rotation motions. Hence, the translation error will affect the heading estimations of the mobile robot, and the accuracy of the FastSLAM algorithm will be reduced over the time. In addition, mobile robots with traditional drives are not suitable for narrow, crowded indoor environment with lots of obstacles. In this paper, an improved FastSLAM algorithm based on Omni-directional wheeled mobile robot is proposed. Computer simulation results demonstrate the effectiveness of the method. This paper provides a methodology to solve the indoor SLAM.

A new method for robot path planning based artificial potential field

The artificial potential field method is used in mobile robot path planning extensively because of its simpleness, high efficiency and smooth path, but it also has its disadvantages. To overcome the shortcomings of the traditional artificial potential field method in mobile robot path planning, this paper analyzes the reasons that lead to the failure in path planning and puts forward an improved method, in which the attractive and repulsive potential field is optimized, also we propose a strategy of potential field filling to escape the GNRON and local minima problems. At last, we introduce regression search to optimize the path. As a result, the mobile robot can find a better and collision-free path to the goal. The simulation result proves the efficient and flexibility of our new APF.

An improved FastSLAM using resmapling based on particle swarm optimization

FastSLAM is well known as a Rao-Blackwellised particle filter formulation of simultaneous localization and mapping. The accuracy of conventional FastSLAM degenerates over time due to the particle depletion in resampling phase. In this paper, an improved FastSLAM with resampling based on Particle Swarm Optimization (PSO) is proposed to address the problem. Firstly, instead of rejection and replication, PSO is employed in the resampling process for the pose convergence of the particle set. Then, in order to update the proposal distribution and the map, a special framework of FastSLAM is adopted in the proposed method. Finally, the result of computer simulation reveals that the modified method shows smaller error in both robot pose and feature estimations than conventional FastSLAM.