Jingzhou Song

3D augmented reality teleoperated robot system based on dual vision

Abstract A 3D augmented reality navigation system using stereoscopic images is developed for teleoperated robot systems. The accurate matching between the simulated model and the video image of the actual robot can be realized, which helps the operator to accomplish the remote control task correctly and reliably. The system introduces the disparity map translation transformation method to take parallax images for stereoscopic displays, providing the operator an immersive 3D experience. Meanwhile, a fast and accurate registration method of dynamic stereo video is proposed, and effective integration of a virtual robot and the real stereo scene can be achieved. Preliminary experiments show that operation error of the system is maintained at less than 2.2 mm and the average error is 0.854 7, 0.909 3 and 0.697 2 mm at x, y, z direction respectively. Lots of experiments such as pressing the button, pulling the drawer and so on are also conducted to evaluate the performance of the system. The feasibility studies show that the depth information of structure can be rapidly and recognized in remote environment site. The augmented reality of the image overlay system could increase the operating accuracy and reduce the procedure time as a result of intuitive 3D viewing.

Study on the perception mechanism and method of virtual and real objects in augmented reality assembly environment

Augmented reality assembly, is a promising application of virtual reality in design and manufacturing and has drawn much more and more attentions from industries and research institutes. The foucs of this study are as follows. (1) using the agent-oriented modeling for subassembly object and the behavior reasoning based on petri net model of environment, to build the system model of augmented reality assembly environment and construct the percepting mechanism of the virtual and real objects; (2) based on above percepting mechanism, to solve the epipolar geometry constraint by least robustness square estimation model, and to sense the depth of the occluded virtual objects or real objects in the environment through assembly scene recognizing, background eliminating and geometrical features restructuring of product contour; (3) combining awareness process of object features matching, constrained motion and features fitting with image based visual hull generating and volume querying, to detect the collision of virtual and real objects, including the dynamic deformable objects. The reseach results is of very important academic and applicative interest in the field of digital design, assembly and maintenace. It would be helpful to promote the application research of augemented reality and virtual reality technology.

Dynamic Obstacle Avoidance Algorithm for Redundant Robots with Pre-selected Minimum Distance Index

A pre-selected minimum distance index is presented considering the real-time performance of dynamic obstacle avoidance for redundant robots.The OBB(oriented bounding box) is applied to modeling robot,and safe links are removed by intersection test before calculating real-time distance.The real-time minimum distance is calculated by local mapping based distance calculation method according to the coordinates of objective points.Based on that,avoidance factor and escaping velocity related to minimum distance are built and then mapping matrix in zero space of redundant robots is utilized to accomplish obstacle avoidance path planning.Finally,the validity and efficiency of pre-selected minimum distance index for dynamic obstacle avoidance are manifested by simulation.Compared with traditional path planning methods,the overall planning time is reduced by 15.5%.

A Novel Design of A Two-Wheeled Robot

According to the theory of spherical robot which invented by our laboratory, we designed a novel designed two-wheeled robot It possesses many salient advantages when compared to traditional mobile robots for its special characters, including self-balance, anti-overturn, and the rests. Its primary driving-method and structure-design arc analyzed and optimized in this paper. Moreover, in this paper, the kinematics model has been established to analyze the moving track of the two-wheeled robot. In addition, the dynamics model based on the circular-plate model has been built to analyze the climbing capacity and the obstacle performance, so as to get the optimum ratifications of the parameters to improve those abilities. Besides, we have manufactured a prototype of this two-wheeled robot, and do some experiments to test the function of this robot, including the maximum speed, the fixed touted movements, and the capability of climbing. And the result coincides with the designed aims.

Urban scene recognition by graphical model and 3D geometry

Abstract This paper proposes a simple and discriminative framework, using graphical model and 3D geometry to understand the diversity of urban scenes with varying viewpoints. Our algorithm constructs a conditional random field (CRF) network using over-segmented superpixels and learns the appearance model from different set of features for specific classes of our interest. Also, we introduce a training algorithm to learn a model for edge potential among these superpixel areas based on their feature difference. The proposed algorithm gives competitive and visually pleasing results for urban scene segmentation. We show the inference from our trained network improves the class labeling performance compared to the result when using the appearance model solely.

Study on real-time interaction method to computer graphics clusters for virtual manufacturing

The paper describes a framework for time-critical rendering of graphics scenes composed of a large number of objects having complex geometric descriptions. An algorithm for mesh simplification both view-independent and view-dependent for original complex model was presented in this paper, First this paper uses edge collapse method with discrete curvature error metric to construct multi-resolution models, then simplified models were provided to view-dependent progressive transmission method to deal with, and implement it with a immediate-mode/retained-mode hybrid distributed parallel rendering system for virtual manufacturing; While this paper use the algorithm called geometry indexing that avoids the redundant geometry transmission by indexing geometry data, alleviating network bandwidth bottleneck.

Projectile motion aerodynamic parameter identification and simulation

The drag coefficient and Magnus coefficient are different along with the geometric parameters of target object when calculating and simulating the trajectory. Generally, the coefficients are measured in wind-tunnel via repeated experiments, and only for object which have the same shape and geometric parameters. This paper proposes a numerical iterative algorithm based on hierarchical optimization method. First, all feasible solutions are calculated with the cost function of carry distance. Then a gradient cost function was added to get the optimal coefficient. In the experiment section, golf projectile motion was employed to verify the algorithm. The result shows that the algorithm is low cost and low error, and just relies on the carry distance of different initial velocity, is more versatility in application.

Research of a New 6-Dof Force Feedback Hand Controller System

The field of teleoperation with force telepresence has expanded its scope to include manipulation at different scales and in virtual worlds, and the key component of which is force feedback hand controller. This paper presents a novel force feedback hand controller system, including a 3-dof translational and 3-dof rotational hand controllers, respectively, to implement position and posture teleoperation of the robot end effector. The 3-dof translational hand controller adopts innovative three-axes decoupling structure based on the linear motor; the 3-dof rotational hand controller adopts serial mechanism based on three-axes intersecting at one point, improving its overall stiffness. Based on the kinematics, statics, and dynamics analyses for two platforms separately, the system applies big closed-loop force control method based on the zero force/torque, improving the feedback force/torque accuracy effectively. Experimental results show that self-developed 6-dof force feedback hand controller has good mechanical properties. The translational hand controller has the following advantages: simple kinematics solver, fast dynamic response, and better than 0.05 mm accuracy of three-axis end positioning, while the advantages of the rotational hand controller are wide turning space, larger than 1 Nm feedback, greater than 180 degrees of operating space of three axes, respectively, and high operation precision.

Manipulability analysis of a two-link space robot using differential geometry method

Space robot is a special robotic system which is expected to perform important tasks in space, like servicing satellites. But any motion of robotic manipulator will disturb its supporting vehicle in space due to the dynamic coupling. Moreover, the evaluating method of manipulability used on ground can not be directly applied to the space robot. A volume element concept is developed to evaluate the manipulability and disturbance of a space robot system. This paper shows the application of volume element method in two-link planar space robot. The volume element method is a new theoretical approach for the research and analysis of space robot system.

Research on real-time geometry compression method in distributed parallel rendering system for virtual manufacturing

The paper describes a mehod for compression of geometry data stream in distributed parallel rendering systems. First this paper uses edge collapse method with discrete curvature error metric to construct multi-resolution models, and use variable quantization of precision to compression 3-dimensional models; then models were provided to view-dependent progressive transmission method to deal with, and implement it with a distributed parallel rendering system for virtual manufacturing; While this paper use the parallelogram prediction of vertex coordinates and the angular space within one sextant of normal to compress 3-dimensional models, alleviating network bandwidth bottleneck.