Y.F. Li

A new camera calibration method from vanishing points in a vision system

The problem of finding intrinsic parameters of a camera is extremely important for practical applications. In this paper, we present an approach for automatic estimation of intrinsic parameters from images by using the vanishing points in orthogonal directions. Firstly, image lines are extracted and clustered into groups corresponding to three dominant vanishing points. Secondly, the intrinsic parameters of the camera are determined by using the vanishing points in each image. The rotation matrix of the projection matrix is computed from the vanishing points and image edges while the translation matrix is obtained with additional translation motion between the viewpoints. Our approach does not need any prior information about the cameras being used. Experiments with real images show that our new technique is robust and accurate.

Incorporating contact sensing in virtual environment for robotic applications

Many robotic tasks involve dynamic interaction between the work piece and handling device, in which sensory information constitutes an indispensable part of the task. It is therefore of practical interest to incorporate sensor modules in the simulation stage to provide a more realistic preview of the operation and predict the behavior of the components prior to the implementation on a real work cell. Most of the current virtual reality (VR) systems use geometric models and sensorless robots for simulation, which cannot provide the insight into the real behavior of the work cell components. In this paper, we report our work in studying the issues related to simulated sensors for dynamic interaction in virtual environment (VE) using physics based models. Based on previous development, a simplified force sensor is modeled and simulated. The current implementation was carried out on a PC based VR platform, with the low-level programs developed using C via the systems development interface (SDK).

Regularization-based recovery scheme for inverse dynamics of high-speed flexible beams

This paper addresses the regularization-based recovery scheme for the inverse dynamics of a flexible beam system. The regularized least square method is introduced in construction of the approximation solution of the inverse dynamics. The well-defined preshaping filter is established based on the regularized least square method for recovering the inverse dynamic solutions satisfying the torque constraints. A numeric simulation is presented to illustrate the validity of the regularization-based recovery scheme proposed in this paper.

Tip deflection measurement and state observation of a flexible robot arm

The knowledge of position, velocity, and acceleration at the tip of a robot arm is useful for advanced control of robots, which is especially true for a flexible robot arm. In this paper, a sensor system is presented to measure in real-time the structural deflections at the tip of a flexible robot arm. This measurement provides a measurement for the tip state observation. A dynamic model of a flexible arm is developed by using the Hamiltons principle. A state space model is obtained with all the state variables having their physical meanings. Based on the dynamic model, an effective state observer is developed by employ the Kalman filter. Numerical simulation results are given to show the effectiveness of the observer.

Metric measurements of 3D object features by a moving camera

With a partially calibrated stereo or shape from unknown motion method, the reconstruction of an object can be performed but at most up to a scale factor. Therefore, metric measurements of the 3D feature cannot be achieved. This paper proposes a method to enable metric measurements of the 3D features starting from 3D length measurement by moving camera. The best focused location in an image is used as a cue in upgrading the relative geometry of a target feature into an absolute one. Hence, we can obtain dimension and metric sizes of object features.

A view planning method for automatic 3D modeling based on the trend surface and limit region

In this paper, a new view planning algorithm of generating 3-D models automatically is presented. It is realized based on the trend surface of the visual field and limit visual region of the vision system. The candidate next best view (NBV) position is obtained by computing the unknown space area according to the trend surface of the acquired the object surface information and limit region of the vision system between viewing points. The final NBV position, which acquired maximum the unknown space area, is got by comparing the above candidate view position.

An approach to dynamic grasp stability measurement

Grasping dynamic stability is an important quality index for robotic grasping. Previous studies on the grasping stability usually only consider the problem under the quasistatic assumption. This paper systematically investigates the grasping dynamic behavior from both theoretical and simulation aspects for the purpose of practical measurements. The dynamic stability of a grasping system is related to the contact type between the grasped object and the fingers, the grasping configuration, the control law of the fingers, and the passive compliance of the fingers. Starting from the dynamic equations of the grasping system, Lyapunov stability theory is used to study the grasping dynamic stability problem, The simulation results show the effectiveness of the formulation for measurements on real systems.

Partial occlusion detection of object boundary

Partial occlusion is a difficult problem in computer vision since whether the object is changed or occluded is ambiguous, especially when distinguishing it only from the object boundary. In this paper, we proposed a novel idea to solve this problem by taking shape matching as a morphing processing. A mass-spring model is constructed from the point set which is sampled from a template (or reference) object boundary by moving it to a target object which is deformed and/or occluded. From the morphing processing, sufficient information can be obtained and an accurate detection of occlusion is performed. By using of the proposed method, the application scope of occlusion detection is expanded while other method cannot be performed which need color, texture, or motion information. The experiments performed on synthetic and real world images proved the satisfactory performance of the proposed method.

View Planning Method Based on the Visual Region of the Vision System

Automatic reconstruction of unknown 3-D objects has been of great importance in computer vision, pattern recognition and visualization. In this paper, a new view planning approach of generating 3-D models automatically is proposed. The new algorithm incorporates the visual region of the vision system with the limit visual surface of the unknown model. The limit visual surface is modeled according to the knowledge of acquired object model. And the space positions of efficient viewpoints are determined according to the visual region of the system and the limit visual surface. The NBV position is the position that could obtain the maximal unknown surface area among the efficient viewpoints. The experimental results show that the method proposed is effective in practical implementation.

Symmetric dichotomy based model of internal dynamics for a class of flexible manipulators

In this paper, we study the symmetric acausal/causal dichotomy for the stable internal dynamics of a class of flexible manipulators. The symmetry computation of the input-state trajectories is also addressed.