Xiaojie Chai

Visual navigation of an indoor mobile robot based on a novel artificial landmark system

This paper presents a practical topological navigation system for indoor mobile robots, making use of a novel artificial landmark which is called MR (Mobile Robot) code. This new kind of paper-made landmarks can be easily attached on the ceilings or on the walls. Localization algorithms for the two cases are given respectively. An approaching control algorithm and an extended line tracking algorithm are also described, which a robot employs to approach its current goal. A simple topological navigation algorithm is proposed. Experiment results show the effectiveness of the method in real environment.

A fast 3D surface reconstruction method for spraying robot with time-of-flight camera

In this paper a new fast 3D surface reconstruction method for spraying robot using TOF (time-of-flight) camera is proposed. The TOF camera can provide 3D points defined in the sensor coordination directly in real time. While fixed on the robot, the TOF camera position and orientation can be estimated and the 3D points can be translated to world coordinate system. An improved fast data acquiring and pre-processing method with TOF camera is proposed to get the point cloud accurately and quickly. In order to obtain 3D faces with given accuracy tolerance from scattered point cloud data rapidly, we present a fast grid building and face extracting algorithm based on flexible cubes. Region segmentation is also need to divide the model to several planes, which making the following path-planning of spraying robot easier. Finally experimental software called Quick RE with friendly man-machine interaction has been developed and the experiment results demonstrate the accuracy and robustness of the proposed algorithm in factory.

Zone divide of three-dimensional model for spraying application on industrial robot

Basing on three-dimensional model, we propose a zone divide method which is suitable for spraying application on industrial robot. The method can divide a model into several meaningful zones by means of smart clustering, the divided zones are split or combined according to the rules designed to improve work efficiency. What is more, in consideration of the specific characteristic of spraying work on industry robot, edge detection imposed on zone is adopted to reduce collision between robot and object when it comes to later real spraying.

Rapid generation of spraying instructions for painting robot basing on automatic programming technology

Basing on automatic programming technology, a rapid spraying instructions generation algorithm is proposed in this paper. Given a patterned three-dimensional model of a workpiece, the algorithm is capable to generate the robot controlling instructions efficiently. On the basis of the above instructions, the industrial robot can paint specified pattern on the workpiece effectively. Innovatively, a zone based path planning algorithm is proposed, which can be used to reduce the complexity of analysis to a great extent. In addition, a separating architecture is put forward, which can be used to improve the versatility of the proposed algorithm effectively. Experimental results show the effectiveness of the method in real environment.

Fast vision-based object segmentation for natural landmark detection on Indoor Mobile Robot

In this paper a new fast vision-based Object Segmentation technique by extracting straight line features from the indoor scenes is proposed. An indoor space scene always contains natural structures like doors, walls, ceilings and floor which have clear straight lines and large homogeneous color surfaces that can be stably detected to form the objects. The objects bounded with lines are very suitable for Indoor Mobile Robot to quickly detect, save as natural landmarks and use in visual SLAM. Compared with the POI (point of interest) features like Harris corner, the line features not only are more robust to changes of scale and illumination, but also can provide more structural information of the indoor environment. This algorithm works in real time and is stable against variation of illumination. The main idea of the method is combining straight lines to form lots of convex polygons. Polygons with homogenous color are kept and adjacent polygons with similar color are merged by a merge test process. A fast line segmentation and fitting method is proposed to improve the line detection efficiency and half edge structure is added to simplify the polygon generation process. Finally experiment results demonstrate the accuracy and robustness of the proposed algorithm in real indoor environments.

Fast selection of face regarding the model in automatic programming techniques

In this paper, based on AABB bounding box[1], spatial subdivision[2], algorithm on whether or not a point is in a certain polygon[3,4], we propose a unique searching strategy. By the aid of our raised method, a fast selection of face in the three-dimensional model can be realized. On that basis, the generated paths can be modified according to the requirements of production process quickly, which extends the range of application for automatic programming[5]. Experiment and result have demonstrated the validity of the proposed method.

Motion planning for vision-based stevedoring tasks on industrial robots

This paper proposed an autonomous robot motion planner for industrial robots with a focus on vision-based stevedoring applications. The planning algorithm can be divided into two stages. The first stage generates initial geometric paths in the Cartesian space: with the 3D model of the environment and the picking and placing pose of the robots wrist obtained by visual system, the planner finds a collision-free path using workspace cell decomposition. The second stage searches for a time-jerk optimal joints trajectories: the planner transforms the path nodes described in the Cartesian space into joints angels in the joints configuration space, then formulates and solves the optimization problem by means of cubic splines. The simulation experiments show the obvious improvement of our method with a contrast to several state-of-art algorithms in this field. And the grasping experiment verifies the practicability and effectiveness of the method on the basis of the embedded visual system and ABB120 type industrial robot.

Path generation for scanning large-size equipment using industrial robot

In manufacturing industry or equipment painting fields, the equipment may have minor shape changes. In order to obtain a better fabrication or painting results, it is required to get an accurate on-site 3D data of the equipment rapidly. A new path generating method on industrial robot with 3D scanner is proposed in this paper. The main idea of the method is to obtain a coarse model of the equipment from the CAD and translate it into the robot coordinate system. Then the model is segmented into flatness zones. Meaningful zone information and collision detection are used to generate the scanning path which could be used by the robots directly. The simulation and the on-site experiment results show its effectiveness and feasibility using the ABBs 5400 robot and TOF depth camera.

An improved visual SLAM algorithm based on mixed data association

Making use of a novel artificial landmark which is called MR (Mobile Robot) code, on the basis of analysis of the motion model and observation model, an improved visual SLAM algorithm based on mixed data association is presented, which improves the localization precision of the robot and the map accuracy. Experimental results verify the effectiveness and robustness of the algorithm.

Extended strong tracking filter SLAM algorithm

Simultaneous Localization and Mapping (SLAM) is a key issue in robotics community. This paper presents a monocular vision and odometer based SLAM algorithm, making use of a novel artificial landmark which is called MR (Mobile Robot) code. During robot motion, the information from visual observations is fused with that from the odometer by Extended Strong Tracking Filter (STF), which can construct highly accurate maps and locate the robot more accurately than EKF. A new calculation method of suboptimal multiple fading factors is proposed which overcomes the problem of discontinuous observation in normal STF SLAM. Actual experiments are carried out in indoor environment, which shows that the proposed algorithm has improved the localization precision of the robot and the map accuracy.