By using the concept of gauge function, a pseudodistance function is defined for quantifying the clearance or the penetration depth of two convex point sets, depending on whether they separate or intersect. The linear programming formulations (for convex polyhedra) and the nonlinear constrained optimization formulations (for general convex objects) are presented for its calculation. For a pair of convex polyhedra, the pseudodistance function is differentiable almost everywhere with respect to the coordinate vectors of their vertices. Sufficient conditions for the differentiability and the characterization of its derivative are presented. By applying the pseudodistance function to the wrench space, a numerical measure of multifingered grasps is defined, which can be used for qualitative test and quantitative analysis of the force-closure property. On this basis, two algorithms for planning optimal force-closure grasps on general three-dimensional objects are developed. In addition, the application of the pseudodistance function in robot path planning is also demonstrated.

A macro-micro manipulator system, which is composed of a long flexible-link manipulator as a macro manipulator and a relatively smaller rigid manipulator attached to the tip of the macro one as a micro manipulator, is employed to improve the motion precision of a flexible space manipulator. At the same time the macro-micro manipulator system preserves the capabilities of large work volume, high operation speed and low energy consumption of the flexible space manipulator. In this study, using the forward/inverse kinematics and a perturbation principle, an error compensation approach is proposed to eliminate as far as possible the endpoint errors of a flexible system through the fast and fine adjustment of the motion of the micro manipulator. A feedback model-based control scheme is proposed to realize the endpoint trajectory tracking of the flexible macro-micro manipulator system. Experimental results on a planar flexible 2-DOF macro/2-DOF micro manipulator system are presented herein to show the effectiveness of the proposed methods.

This paper presents a new approach for fitting a subdivision surface from an irregular and dense triangle mesh of arbitrary topological type. All feature edges and feature vertices of the original mesh model are first identified. A topology- and feature-preserving mesh simplification algorithm is developed to further simplify the dense triangle mesh into a coarse mesh. A subdivision surface with exactly the same topology and sharp features as that of the simplified mesh is finally fitted from a subset of the vertices of the original dense mesh. During the fitting process, both the position masks and subdivision rules are used for setting up the fitting equation. Examples are provided to demonstrate the proposed approach.

In this paper, a new adaptive coupling control algorithm is developed to synchronize motions of two driving wheels of a differential mobile robot so that the robot maintains in the desired trajectory path. Unlike considerable approaches in the literature, in this study motion of driving wheels instead. of the robot configuration are directly controlled in a synchronous manner. This strategy avoids the involvement of nonholonomic constraint in the control design, and therefore simplifies the practical implementation significantly. The proposed controller incorporates the cross-coupling technology into adaptive control architecture, and guarantees asymptotic convergence to zero of both position tracking error and the synchronization error. Experiments have been conducted to verify the effectiveness of the proposed control approach.

A new control algorithm for tracking stabilization of a differential mobile robot is developed by incorporating the cross-coupling technology into an adaptive control architecture. Unlike the majority of previous work in the literature, coordinates of robot wheels instead of the robot configuration, are directly controlled in a synchronous manner. The synchronization effort is aimed to regulate the differential position error of two driving wheels to zero and thus regulate the heading angle of the robot, so that the robot maintains in the desired trajectory path. The proposed adaptive synchronized controller guarantees an asymptotic convergence to zero of wheel displacement errors and the synchronization error between them. Under the condition that the initial error of the robot configuration is zero, the robot configuration converges to the desired state asymptotically. Experimental results demonstrate the effectiveness of the proposed approach.

Although many medical robots have been developed, such as surgical robots, nurse robots, etc., very few robots are introduced in the research field of oral restoration, especially false-teeth restoration. Currently, false-teeth restoration is done manually. Robotic manufacturing system of complete dentures is described in this paper, which performs a complete false teeth automatic implanting. It relies on the use of a special light sensitive material that hardens under lighting. In this system, a robot grasps selected standard teeth and implants them in fixed positions. Expert system software is developed which integrates high-level dentist's experience and technician's dexterous skill, it creates the control data for the robot to implant every tooth correctly into the denture base. The software can display 3D pre-composition of complete denture and realized man-machine interaction. The dentist can easily and conveniently adjust and modify every tooth's position and orientation in complete dentures. This software can also realize path planning and data communication between the computer and the robot controller. This robotic system will change the manner of complete denture manufacturing, significantly improving its quality efficiency.

Considers the tracking problem of dynamic nonholonomic systems with unknown inertia parameters. A new controller is proposed. The proposed controller not only ensures the entire state of the dynamic system asymptotically track the desired trajectory, but also is characterized by low dimension and the absence of singular points. Simulation results show that the proposed controller is effective.

Wavelet transforms and fuzzy techniques are used to monitor tool breakage and wear conditions in real time according to the measured spindle and feed motor currents, respectively. First, continuous and discrete wavelet transforms are used to decompose the spindle and feed ac servo motor current signals to extract signal features so as to detect the breakage of drills successfully. Next, the models of the relationships between the current signals and the cutting parameters are established under different tool wear states. Subsequently, fuzzy classification methods are used to detect tool wear states based on the above models. Finally, the two methods above are integrated to establish an intelligent tool condition monitoring system for drilling operations. The monitoring system can detect tool breakage and tool wear conditions using very simple current sensors. Experimental results show that the proposed system can reliably detect tool conditions in drilling operations in real time and is viable for industrial applications.

A simple general motion planning approach for nonholonomic mechanical systems of chained from is proposed, in which the intricate motion planning problem (steering the system from an initial state to a final state) is converted to a simple curve fitting problem (satisfying a set of end-point conditions). By means of this approach, other geometric constraints can be easily handled, while the control inputs can be derived directly from the smooth path planned. For verifying the effectiveness of the proposed approach, a number of simulations are conducted with respect to a three-input firetruck (two-chain system).

Fuzzy reactive control, incorporating a local target switching scheme, is applied to the automatic navigation of an intelligent mobile robot in an unknown and changing environment. Sensed-ranging signals and relative target position signals are input to the fuzzy controller. The steering angle and the velocity change are inferred to drive the mobile robot. A reactive rule base governing the robot behavior is synthesized from the human heuristics with respect to various situations of environment. A local target switching scheme is proposed to serve as a front-end processor of the fuzzy active controller and to deal with the local trapping and wandering cycle problem in the navigation of a behavior based mobile robot. The algorithm is described, together with some particular considerations about implementation. Efficiency and effectiveness of the proposed approach are verified through simulation and experiments conducted on a Nomad 200 mobile robot.

Linear controllers perform well for straight-line path-following autonomous vehicles. For curved paths, significant large errors may result. These can be improved using two linguistic fuzzy-logic controllers (FLCs) even in the presence of inaccurate measurement of path curvature. One of the FLCs is used for accounting the path curvature. Using the linear-to-fuzzy algorithm, the synthesis of the linguistic FLCs is readily achieved. Also using the equivalent transformation algorithm, the stability analysis of the linguistic FLCs can be carried out based on a Lyapunov function which is obtained using the linear-matrix-inequality method. Experimental results are presented to illustrate the effectiveness of the proposed FLCs.

This paper proposes a new demonstrated-trajectory modelling technique for human-skill representation based on differential-geometry properties. It satisfies three desirable criteria for generalization of the captured human trajectory for trajectory comparison and robot-command mapping. The important criteria are spatial independence, temporal independence, and suitability for similarity measurement. The differential geometry properties extracted for use are curvature and torsion which adequately describe the trajectory shape and fulfil the first two criteria. The last criterion is met by introducing a suitable form of model function derived from the curvature and torsion functions.

A virtual target approach is proposed for resolving the limit cycle problem in navigation of a behaviour-based mobile robot. Starting from the onset point of a possible limit cycle path, the real target is switched to a virtual location and the robot is navigated according to the virtual target set up temporarily and the real environment information sensed, until a switching back condition is reached. The condition for switching back to the real target is established using a specific change in the obstacle information sensed. The algorithm is described together with some particular considerations in implementation. Efficiency and effectiveness of the proposed approach are verified through simulation and experiments conducted with a Nomad 200 robot incorporating a fuzzy, behaviour-based controller.

This paper proposes an automatic selection scheme to choose the most consistent trajectory among a number of human-demonstrated ones. The consistency-determination is based on the hidden Markov model (HMM) technique. There are three stages involved. The first stage is preprocessing of the human-generated trajectories. It includes short-time Fourier transform and vector quantization. The former maps the trajectories from the time domain to the frequency domain, and the latter quantizes a list of frequency spectra to a finite number of prototype spectrum-vectors, called symbols. The second stage is training of the HMM. The unknown model parameters in the HMM are tuned by the concept of counting event occurrences. The quantized symbols are counted so that probabilities of occurrences are applied to train the HMM. The third stage is measurement of the consistency of every trajectory. Each trajectory is sent through the trained HMM. A probability-based likelihood index is evaluated which reflects the consistency of the trajectory with the HMM. The trajectory giving the largest likelihood index is considered to be the most consistent one and will be selected.

It is the function of the design of a fuzzy-logic controller to determine the universes of discourse of the antecedents and the consequents, number of membership labels, distribution and shape of membership functions, rule formulation, etc. Much of the information is usually extracted from expert knowledge, operator experience, or heuristic thinking. It is hence difficult to mechanize the first-stage design of fuzzy-logic controllers using linguistic labels whose performance is no worse than that of conventional multivariable linear controllers such as state-feedback controllers, PID controllers, etc. In this paper, an original systematic seven-step linear-to-fuzzy (LIN2FUZ) algorithm is proposed for generating the labels, universes of discourse of the antecedents and the consequents, and fuzzy rules of ;basically linear' fuzzy-logic controllers, given the reference design of available conventional multivariable linear controllers. The functionally equivalent fuzzy-logic controllers can thus provide the sound basis for the further development to achieve performance beyond the capability or the conventional controllers. The validity and effectiveness of the proposed LIN2FUZ algorithm are demonstrated by a four-input one-output inverted pendulum system.

The performance of automatic guided vehicles (AGVs) is gauged by their position and orientation errors. Sometimes the orientation errors are not measurable, and output feedback using position errors only is employed. In this paper, a controller using output feedback is proposed for a differential-wheel-drive AGV with a simple linear array of 64/spl times/1 opto-sensors. A linear controller is first developed. The performance of the linear controller is further improved using fuzzy logic. Two fuzzy-logic controllers are accordingly formed. Stability based on the Hurwitz criterion, the Popov criterion, and the phase-plane method is demonstrated. Experimental results using a laboratory AGV are provided to illustrate the effectiveness of the proposed controllers.

Autonomous vehicles often move along rather arbitrary types of path each demanding a different strategy for effective control. While it has been shown in previous papers that basic fuzzy (nonlinear) controllers can perform better than linear optimal controllers, it is desirable to take more fuzzy variables into account in order to cope intelligently with a wider variety of situations. By means of actual examples and using a physical mobile testbed, the application of intelligent fuzzy switching is investigated, which reduces a three-variable problem to a two-level lower-order problem, with reasonably smooth blending of the control strategies according to different path conditions, and with significant reduction of design effort used to generate the fuzzy control law. Experimental results are provided to demonstrate the usefulness of the proposed approach for dealing with vehicle paths characterised by nonlinear shapes and abrupt curvature changes.

The paper describes a method of generating robot program codes automatically from perception of human demonstration. The movement of a marker-based mechatronic input device representing the end-effector of the manipulator is driven by an operator, and recorded by a special visual measurement system. The captured spatial path is divided into different segments for individual processing. For the 'pre-grasping' and 'post-degrasping' segments, only the end-effector 6D information during grasping and degrasping is extracted for input to the high-level grasping and degrasping commands. For the 'manipulating from start' and 'manipulating to stop' segments, spatial data are interpolated by the cubic-spline interpolation algorithm to provide sufficient data points for some robot motion controllers. For 'manipulation at high rate' and 'translation at high speed' segments, the segment start and end points are input as the two-point-boundary-value arguments of an optimal controller for generating better trajectories.

The paper describes a visual programming approach to transferring the operator's manipulation skill directly to the robot. The objective is to let the robot learn by observing the human expert moving in his/her own work space. High-speed and high-precision acquisition of the location information is expedited by a marker-based visual system mounted on a conic tool surface, assisted by transputer calculation. Possible violation of the robot physical constraint is checked online with the provision of computer graphic facilities in the system which are also used for subsequent automatic path refinement.

This paper describes a fast motion planning method which can be applied to online collision avoidance. It is based on C-obstacle determination, but the way in which the surface contours are established reduces significantly the determination time. Collision checking is based on exploring possible virtual intersection between the two convex polyhedra which define the robot link and the obstacle. Virtual intersection is deduced by examining the signs of distances between a facet of a robot polyhedron and the vertices of the obstacle polyhedron, and that between a facet of the obstacle polyhedron and the vertices of the robot polyhedron. A subsidiary feature of this planner is that it can determine the direction for the robot to retract from an approaching obstacle, according to the types of virtual intersection detected.

The authors describe an iterative learning controller for robot manipulators. The learning control method presented makes use of partial model knowledge of the robot system and systematically merges desirable types of sensor data for iterative control. Fast convergence can always be obtained without going through complicated tuning exercises. The experimental results showed the superior learning performance of the robot system working with the learning controller.>

Classical and adaptive control of robot manipulators can be either joint-based or Cartesian-based. The formulation schemes of learning control of manipulators subject to repetitive operation described in the literature are apparently all based on the joint space. A new discrete-time formulation of Cartesian-based iterative learning control is developed in which partial knowledge of the robot dynamic model is utilized. From the characteristics and implementation points of view, it appears to be more attractive than existing learning control schemes. The convergence issue is particularly addressed.>

With a view to improving the quality of learning and simplifying the iterative learning controller implementation, a new discrete strategy for robot motion control is proposed. A systematic means is devised to merge the position and velocity data in consecutive cycles for effective learning control. The convergence property is also established.>