Didgeridoo

“Summarized” history of FlightGear

This project started of a discussion between Net surfers in 1996, which a proposal written by David Murr left, who gave up the project (and the Net) later. The original proposal is still available on the Web site of FlightGear and can be found with http://www.flightgear.org/proposal-3.0.1

Although names of people and some details changed since time, the spirit of this proposal was clearly kept until now.

Current coding started at the summer 1996 and at the end of this year, the essential graphic routines were written. At that time, the programming was mainly carried out and coordinated by Eric Korpela of the university of Berkeley. Very early, the code turned under Linux as well as under DOS, OS/2, Windows 95 /NT and Sun-OS. It was a rather ambitious project which required, inter alia, to write all the graphic routines independent of the system entirely from nothing.

The development slowed down then finally stopped at the beginning of 1997 when Eric finished his thesis. At this time, the project seemed dead and the number of messages on the list of discussion was close to 0.

It was Curt Olson of the university of Minnesota which started again the project in the middle of 1997. Its idea was as simple as strong: why invent the wheel one second time? There are several free flight simulators available on work station under various Unix systems. One of them, LaRCsim (developed by Bruce Jackson of NASA), seemed to be perfect in this approach. Curt took this one separately and rewrites several routines so that it can be built and carried out on the target platforms envisaged. The key idea was to exploit a graphic platform independent of the system: OpenGL.

Moreover, an astute decision for the selection of the data of the basic scene was made in the very first versions. The scene of FlightGear is created on a database satellites published by the U.S. Geological Survey. These data of grounds are available for the whole world on Internet free to http://edcwww.cr.usgs.gov/doc/edchome/ndcdb/ndcdb.html for the United States, and http://edcwww.cr.usgs.gov/landdaac/gtopo30/gtopo30.html, for the other countries. These free data, in conjunction with the tools for manufacture of scenes included in FlightGear, are an important possibility making it possible whoever to create its own scene.

This new code of FlightGear - still largely based on the original code of LaRCsim - was delivered in July 1997. As from this moment the project still saved time. There were big steps in the more recent history of the development:

the posting of the sun, the moon and stars is a possibility in which the simulators for PC are manifestly weak since years. It is one of the great successes of FlightGear which to include an exact model and the posting of the sun, the moon and planets. The astronomical code correspondent was added in autumn 1997 by Durk Talsma.

textures were added by Curt Olson to spring 1998. That marked an significant improvement in terms of realism. You will note that Microsoft Flight Simulator had scenes not textured to version 4.0. Some textures of high-quality were provided by Eric Mitchell for the FlightGear project.

a Affichage head high (head up display - HUD) was added, based on the code of Michele America and Charlie Hotchkiss in autumn 1997 and was improved by Norman Vine. Although it is usually not available in reality for Cessna 172, the HUD is practical to post information on simulation in progress. However, the instruments behave as on the real aircraft (including with their errors) and that can be very destabilizing for a pilot apprentice or a developer.

After having improved the scene and textures and addition some other possibilities, the frequency image (“frame spleen”) fell at the point to return FlightGear involable in spring 1998. The solution was to use the material part of the OpenGL charts, which became available to this date; to program “view frustrum culling” (a technique of visualization which is unaware of the nonvisible part of the scene), this made by Curt Olson. These measurements made FlightGear usable again - even on not very powerful machines - since a graphics card 3D with the support of the OpenGL hardware is in the machine. With regard for the frequency of images: one must keep in mind that the code, now, in no way is optimized, which leaves the possibility for improvements in the future.

a rudimentary autopilot (maintenance of the course) was the contribution of Jeff Goeke-Smith in April 1998. It was improved with the addition of the maintenance of altitude and a “ground following switch”, in October 1998.

the bases of a system of menu was brought by portable library PLIB of Steve Baker in June 1998. After having been unutilised a time, the first menus were active in spring 1999.

PLIB underwent a rapid development then. It was distributed by Steve in a separate packet with the idea to be useful to other applications since spring 1999. It provides the bases of an engine of made graphic for FlightGear since the autumn 1999.

Friedemann Reinhard early developed the code of an instrument panel, which was available in June 1998. The development of this table stopped then, partly because of compatibility issues with OpenGL. Finally, David Megginson decided to set out again zero in January 2000, the result being that the table is now again under fast development.

In 1998 there was the basic audio support, i.e. a library (of routines) and some sounds of engines. It was the contribution of Steve Baker and it then was integrated in its already mentioned portable library, PLIB. This same library was extended to support the handles of play, wheels, swing bars in October 1999, which still marked a big step in realism.

In September 1998 Curt Olson succeeded in creating a model of complete ground for the the United States. The scene is available for all while clicking on a chart to http://www.flightgear.org/Downloads/scenery.html

The scene was improved by adding geographical reference marks like the lakes, rivers and coasts in spring 1999.

the code for the multijoueurs/network management was integrated by Oliver Delise and Curt Olson into the beginning of the autumn 1999. This effort made it possible FlightGear simultaneously to function on several machines through a lan, in Intranet or on Internet; coupled to a planner of flight turning on one second machine, and more.

Christian Mayer and Durk Talsma, added the management of the weather during the winter 1999. That manages the clouds, the winds and even the storms.

Changer manually the sights in a flight simulator is in a certain “unreal” direction always but nevertheless necessary in unquestionable situations. A possible solution was provided by Norman Vine in winter 1999. Norman wrote the code to change sight with the mouse.

Finally, Navion of LaRCsim was replaced like plane by defect by Cessna 172, when this one became rather stable in February 2000 - a welcome change for the majority of the users. There are now several options of models of flight to choose with the execution: Cessna 172 with modified and improved LaRCsim developed by Tony Peden, X15 de Jon Berndt, and the balloon with hot air of Christian Mayer. Jon Berndt invested much time in model of a more realistic and flexible flight, with a method of more powerful configuration of the planes. JSBSim, as it was going to be called, can possibly replace LaRCsim like dynamic model of flight (FDM - Flight Dynamics Model) and it is envisaged to add to it some functionalities like the effects of clapotage (NDT?) fuel, turbulence, a control system of complete flight, and other things seldom found together in a simulator of vol.

During the development, there were several efforts of reorganization of the code. Subsystems of varied codes were put out of packages. Currently, the code is organized as follows:

The user must have a video chart 3D -- preferably one with material management of OpenGL. Based on OpenGL, portable library PLIB of Steve Baker provides the basic routines of made graphic, the management of the sound, the handles of play, etc Basé on the PLIB there is SimGear, which as well as includes all the basic routines necessary for the simulation of the flight for the construction of the scene. Above SimGear there are (I) FlightGear (the simulator itself), and (II) TerraGear, which includes/understands the tools of contruction of scenes.

Since the summer 1999 FlightGear was divided into a stable branch and another of development. Each number of version like 0.6,0.8, and day if all is well 1.0 refers to stable versions, while the odd numbers 0.7,0.9, and so on refer to versions of development. The policy is to make only resolutions of bugs in the even versions, whereas the new functionalities are generally added in the odd versions which, when all is stabilized, become the next stable version with a number calculated by adding 0.1.

It is certainly not a complete history and certain people who made important contributions were probably not quoted. Except the already named contributions there was an important work concerning the internal structure carried out by: Steve Baker, Jon S.Berndt, Oliver Shred, Christian Mayer, Curt Olson, Tony Peden, Gary R. Van Sickle, Norman Vine, and others. A more complete list of the participants can be found in the handbook, and also in the Thanks file provided with the code. The Web site of FlightGear contains also a detailed history of all the notable developments with http://www.flightgear.org/#news/

To note the arrival running November 2004 of a Japanese and French translation of the official site, coming to join, with the German community, the international community of this simulator of vol.

Software

The engine of simulation is SimGear. It is used as much for final applications as in environment of research, or than for the development of simulations of vol.

This fickleness of Flightgear is illustrated perfectly by the large variety of aircraft available, energy of the sailplane to the helicopter, while passing by the airliners and the hunters. These aircraft originate in the Flightgear community.

An aircraft Flighgear 0.9.10 uses one of the three data models JSBSim, Yasim, or UIUC. Currently, only an engine of returned ground is available: TerraGear. The weather effects include the clouds 3D, the flashes during the storms and the current hour.

Models of dynamic flight

The model of flight is the way in which an aircraft in the program is simulated. Flighgear can at present use several models of flight among the three available ones and each aircraft must be programmed specifically for one of these models.

The first versions used a model called LaRCsim, developed by the NASA, which was replaced later by more flexible models:

JSBSim - the model by defect since 2000.
YASim - Another model using of the different algorithms and available since 2002 in the version 0.7.9
UIUC - Another model, developed by the UIUC Applied Aerodynamics Group of the university of Illinois in Urbana-Champaign, which used LaRCsim.

Captures of screen

Random links:

Academic painting | Saint-Vallier (Saône-et-Loire) | Canton of Épernay-1 | Alaskan Klee Kai | Alley Cat