MIT Design Laboratory Current Research

Solution of Nonlinear Systems

A fundamental problem in computer aided design is the robust and efficient computation of all real roots of systems of nonlinear polynomial or more complex equations. Such problems often occur in computing intersections, distance functions, and extrema, as well as in engineering design. In order to isolate all roots within a given domain, several methods are being studied: the first method projects control polyhedra onto a set of coordinate planes; the second employs linear optimization; and the final method utilizes an interval Newton technique. Interval arithmetic implementation is a basic component of these methods leading to computational reliability.

Robust Geometric Modeling

Full automation of the design of complex structures requires reliability of the geometric computations involved in the design process. However, this reliability is not present in current modelers due to floating point operations that lead to inconsistent or unstable numerical computations. The goal of this project is the development of a solid modeling system based on interval splines. Interval splines differ from classical splines in that the real numbers representing control point coordinates are replaced by intervals which in combination with rounded interval arithmetic can guarantee the robustness of geometric computations.

Praxiteles

In order that the design and manufacture processes be interactive, algorithms developed in the MIT Design Laboratory for shape creation, advanced interrogation, visualization, fairing, and inspection have been integrated into an executive system, called Praxiteles. Praxiteles, named after the sculptor of the statue of Hermes in Olympia, was developed for use on the Silicon Graphics IRIS 4D series workstations which provide for graphical output of numerical computations and allows for input and output via the IGES format which provides a standard for the exchange of geometric data with other CAD systems.

In addition, since many systems are limited to cubic piecewise polynomial geometries defined in terms of B-splines (for efficient and stable processing), Praxiteles facilitate the exchange of curve and surface geometry between different geometric modeling systems through advanced methods for approximate conversions of high degree Bezier and B-spline surfaces to lower degree representations with guaranteed bounds on the approximating error.

Advanced Methods for Inspection

Inspection is the process of verifying shape conformance of a measured manufactured part with a toleranced geometric description of a design object. Since manufacturing processes such as milling and casting are inaccurate, the need for accurate inspection methods is fundamental, particularly for objects that require a very high level of precision. Three processes for inspection take the form of feature extraction, which is the ability to extract performance related geometric features from a mathematically represented design or as-built model; of localization, which is the optimal positioning of a mathematically represented design model with respect to a measured manufactured artifact; and of local error analysis, which is used for planning remachining operations.

Idealization for Mesh Generation Using MAT

This project concerns the generation of coarse and fine meshes on multiply connected 2D and 3D domains based on the medial axis transform (MAT). The MAT or skeleton or symmetric axis of a shape is a point set consisting of points that are equidistant from two or more points on the boundary contour. Using the MAT, we can extract some important shape characteristics and their length scales so that we can create a coarse subdivision of a complex surface and then select a local element size to generate fine triangular meshes within individual subregions. The MAT allows us to carry out these processes in an automated manner. Thus, our approach can lead to integration of automated finite element (FE) mesh generation schemes into CAD systems.

Advanced Techniques for Milling Operations

Propeller blades are manufactured by numerically controlled (NC) milling machines. When a ball end-mill cutter is used, the cutter radius must be smaller than the smallest concave radius of curvature of the surface to be machined to avoid local overcut (gouging). Gouging is the one of the most critical problems in NC machining of free-form surfaces. Therefore, we must determine the distribution of the principal curvatures of the surface, which are upper and lower bounds on the curvature at a given point, to select the cutter size. We are developing a robust algorithm for contouring curvature of a free-form surface using our work on the solution of nonlinear equations. We then subdivide the surface into regions of similar order of magnitude curvature which can be used for both tool selection and tool path generation.

Interrogation and Visualization of Non-Linear Databases

The objective of this project is to advance the state of knowledge in efficiently representing, interrogating, visualizing, updating and manipulating vast amounts of geophysical data. Our work involves concepts from interval arithmetic, differential topology, algebraic topology, all applied in computational geometry representations and algorithms. We are currently addressing the problems of multidimensional physical data representation using advanced higher dimensional geometrical concepts and of uncertainty in data representation and computation by invoking the concept of interval n-dimensional B-splines and simplex splines and the theory of interval arithmetic.