Djordje Brujic

Analysis of free form surface registration

Registration of a rigid body is an essential step in many machine vision tasks. This paper presents performance analysis of the iterative closest point method implemented for dimensional inspection of free form surfaces where the set of measured points is registered to a NURBS model. The results were obtained through Monte Carlo simulation. A large number of experiments were performed on an example precision engineering component. Influences of shape, number of measured points and measurement noise on registration accuracy are identified through statistical analysis.

Measurement-based modification of NURBS surfaces

A frequent requirement in computer aided design and manufacture is to update or refine an existing CAD model using measured data. Least squares surface fitting is known to suffer from stability problems, caused by an insufficient measurement density in some regions. This is particularly evident in situations involving local surface updating and when knot insertion is applied for local surface refinement. This paper presents a new method to update the CAD model consisting of NURBS surfaces, trimmed or untrimmed, based on a set of unorganised measured points in three-dimensional space. The proposed method overcomes the fundamental problem of singular or ill-conditioned matrices resulting from incomplete data sets. This was achieved by introducing additional fitting criteria in the minimisation functional, which constrain the fitted surface in the regions with insufficient number of data points. Two main benefits were realised by this approach. First, local surface updating can be performed by treating the surface as a whole, without the need to specially identify the regions with insufficient data, nor to re-measure those regions. Second, the quality of the unmeasured regions may be controlled to suit specific needs. The results were found to be highly encouraging and the method was found to be especially useful in situations involving knot insertion and large surface deformations.

Efficient registration of NURBS geometry

Abstract The paper presents an implementation of free-form surface registration in relation to inspection of engineering components, defined as NURBS. Registration is principally performed through the Iterative Closest Point (ICP) method. The time-critical step in ICP was found to be the determination of the closest points on NURBS to a given point in space. Significant speed improvements were achieved through the adoption of a dual surface representation, involving approximation of NURBS entities by a polyhedral mesh. A criterion for sufficient polyhedral approximation was derived and implemented, producing encouraging results. Original solutions are suggested in order to further improve the computational speed. Extensive testing has been carried out, showing that the proposed registration method handles a full six degrees of freedom and achieves global convergence. Performance of the implemented algorithms is discussed with reference to registration of a turbine blade airfoil.

Economic dispatch of distributed combined heat and power systems participating in electricity spot markets

An optimization tool is proposed to determine the optimal operation schedule for distributed combined heat and power plants which participate in electricity spot markets. It was tested and analysed using realistic data for heat and electricity demands and their corresponding prices. The results show substantial cost savings compared with the benchmarked alternatives.

Measurement-based updating of turbine blade CAD models: a case study

Manufacturing industries, especially the aerospace and the automotive, require updating of CAD models so that the manufactured geometry is accurately represented. Often, additional constraints such as continuity are imposed on the model. The updated model may be employed to provide a more realistic analysis of the parts in-service performance. The necessary tools for realizing this task included dimensional measurement and geometric modelling. This paper discussed how this was achieved in the case of an aeroengine gas turbine blade. Measurement was performed using a coordinate measuring machine equipped with a touch trigger probe. Measurement path plan is based on the pre-processed CAD model. The accuracy was improved using our method for probe radius compensation. The re-modelling steps include preparation of the base surface, registration and least squares surface fitting. The surface preparation is needed in order to maintain prescribed continuity. Also, special attention was paid to cure ill-conditioning, which occurs when fitting trimmed surfaces. The adopted novel solution involves regularization and introduces two additional terms in the fitting functional. Two weighting coefficients are introduced to improve flexibility of the solution and they represent the users confidence in the measurements and quality of the initial model.

Monte Carlo simulation and analysis of free-form surface registration:

Abstract Accurate dimensional inspection and error analysis of free-form surfaces requires accurate registration of the component in hand. Registration of surfaces defined as non-uniform rational B-splines (NURBS) has been realized through an implementation of the iterative closest point method (ICP). The paper presents performance analysis of the ICP registration method using Monte Carlo simulation. A large number of simulations were performed on an example of a precision engineering component, an aero-engine turbine blade, which was judged to possess a useful combination of geometric characteristics such that the results of the analysis had generic significance. Data sets were obtained through CAD (computer aided design)-based inspection. Confidence intervals for estimated transformation parameters, maximum error between a measured point and the nominal surface (which is extremely important for inspection) mean error and several other performance criteria are presented. The influence of shape, number of measured points, measurement noise and some less obvious, but not less important, factors affecting confidence intervals are identified through statistical analysis.

Needle-Guiding Robot for Laser Ablation of Liver Tumors Under MRI Guidance

This paper presents the design, control, and experimental evaluation of a needle-guiding robot intended for use in laser ablation (LA) of liver tumors under guidance by magnetic resonance imaging (MRI). The robot provides alignment of a needle guide inside the MRI scanner bore and employs manual needle insertion. In order to minimize MR-image deterioration, the robot is actuated using plastic pneumatic cylinders and long pipes connecting to control valves located outside the MRI scanner room. A new time-delay control scheme was employed to achieve high position accuracy without requiring pressure or force measurements in the MRI scanner. The control scheme was compared with experiments to a previously developed sliding mode controller. A marker localization method based on the convolution theorem of Fourier transform was employed to register the robot in the MRI scanner coordinate system and to verify the position of the needle guide before the manual needle insertion. Experiments in a closed-bore MRI scanner showed a variation in SNR below 5%. A phantom study indicates that the targeting error in robot-assisted needle insertions is below 5 mm and suggest a potential time saving of 30 min compared to the manual MRI-guided LA procedure.

Contact probe radius compensation using computer aided design models

Abstract Probe radius compensation is necessary in metrology applications that employ contact probes, but it can be a significant source of systematic measurement errors when dealing with free-form part geometry. The paper presents implementation and performance analysis of a proposed new compensation technique based on the nominal computer aided design (CAD) model, which is assumed to be defined using non-uniform rational B-splines (NURBS). Errors associated with the conventional compensation approach are assessed on the basis of experiments using a modern coordinate measuring machine (CMM), providing clear motivation for this work. The proposed method consists of a number of steps, including measurement, generation of offset nominal surfaces, registration, surface fitting, data smoothing and calculation of compensating offsets. Critical steps include registration and NURBS surface fitting and their implementation is presented. Simulation studies are used to analyse the registration accuracy and the accuracy of the overall method in comparison with the conventional one. The proposed method is shown to produce superior results in situations involving non-uniform measurement distribution, measurement noise, free-form geometry with no clear datums, deformation relative to the nominal shape and component misalignment.

Optimal Operation of Distributed CHP Systems for Participation in Electricity Spot Markets

An optimisation tool is proposed to determine optimal operation schedule for distributed combined heat and power (CHP) plants which participate in electricity spot markets. It was tested and analysed using realistic data for a heat and electricity demands and their corresponding prices. The results show substantial cost savings compared to the benchmarked alternatives.

A framework for non-contact measurement and analysis of NURBS surfaces

The paper presents an implementation of a methodology for reliable and feasible dimensional measurement of engineering components containing free-form surfaces. Two main requirements had to be satisfied. First, a large number of points had to be accurately measured. Secondly, the set of corresponding points on the nominal model had to be computed within a reasonable time. The first aspect was satisfied by adopting a non-contact measurement technique based on laser triangulation. High accuracy was achieved through software prealignment for precise component localisation and through appropriate measurement planning, both based on the CAD model. Determination of the corresponding nominal points was solved by best-fitting. Significant speed improvements were achieved through an implementation of the iterative closest point algorithm, based on a dual representation of the surface. The nominal surface is defined using NURBS entities and its approximation is determined as a polyhedral mesh. A sampling criterion for complex surfaces was derived and implemented, producing encouraging results. Validity of the proposed approach is supported by experiments and by simulation studies involving real engineering components.