Predictive look ahead memory management for computer image generation in simulators

• US 4,952,922 A
• Filed: 07/18/1985
• Issued: 08/28/1990
• Est. Priority Date: 07/18/1985
• Status: Expired due to Fees

• Current Assignee: Hughes Aircraft Company (Rtx Corporation)
• Original Assignee: Hughes Aircraft Company (Rtx Corporation)
• Inventors: Griffin, Arthur F., Moise, Norton L.
• Primary Examiner(s): Lobo, Ian J.
• Application Number:
  US06/756,869
• Time in Patent Office: 1,867 Days
• Field of Search: 434/38, 434/41, 434/44, 434/35, 434/37, 364/578, 364/518, 364/423, 364/522, 340/729
• US Class Current: 345/421
• CPC Class Codes: G06T 15/06 Ray-tracing
• CPC Class Codes: G06T 15/40 Hidden part removal
• CPC Class Codes: G09B 9/301 by computer-processed or -generated image

Abstract

A method for use with a vehicle simulator which provides for the rapid update of displayed terrain data in areas of non-visible terrain surfaces and terrain image discontinuities. The method is for use with a system which projects a time-varying computer-generated image scene representative of the terrain as viewed from a moving vehicle. The image scene is derived from a digital data base stored in slow speed secondary memory. The method includes the steps of predicting, by means of line-of-sight sampling ray, which segments of the data base could be visible after a predetermined time interval. The data base is accessed for the predicted segments of data and these segments are retrieved therefrom. The data segments designated are then stored in predetermined memory locations in high-speed main memory. Finally, selected ones of the stored segments are displayed based on the actual vehicle path and position after the predetermined time interval has elapsed.

Claims

1. A method for use in a system which projects a time varying computer generated image scene representation of terrain as viewed from a moving vehicle which traverses a simulated path, said image scene being derived from a data base stored in secondary storage, said method comprising the steps of:

predicting a set of possible future positions of the vehicle after a predetermined time interval, said predicted positions being calculated based on the motion dynamics of the vehicle;

for a number of possible future positions, sampling the terrain data base with line-of-sight rays projected from the position of the vehicle operator'"'"'s eye into the data base in cartesian coordinate space;

marking the data at the point of sample ray terrain surface intersection and the data for surfaces adjacent to that point as likely candidates for near-future display update;

accessing the marked data and storing them in high speed main memory; and

displaying selected ones of the stored data segments based on the simulated path and position of the vehicle after the predetermined time interval has elapsed;

whereby a rapid update of the displayed terrain in the near future is achieved.

2. A method to generate sampling rays as claim 1 by sampling picture elements (pixels) within the field-of-view in the plane of the display and projecting rays from the eye position through each of the sample pixels.

3. A method to store and retrieve scene data, as claimed in claim 1, that is a collection of elements corresponding to cubic volumes that make up a three dimensional terrain model, only those volume elements (voxels) containing terrain surfaces are stored, said method utilizes a bit map of all voxels in the terrain model, each bit having a memory address that is computable from the x, y, z cartesian coordinate of the voxel, the binary state of each bit so addressed indicating whether the corresponding voxel does or does not contain a terrain surface.

4. A method to mark the data to be accessed as claimed in claim 1 by selecting a set of data base volume elements (voxels) that form a cube whose center is the voxel containing a terrain surface intersected by a projected sampling ray, the dimensions of said cube are calculated as a direct function of the distance from the eye position to the intersected surface voxel.