So you think it’s a challenging task to develop an autonomous vehicle run by a sophisticated software program to respond to external stimuli in real time, be adaptive to situations requiring quick response while analyzing a multitude of external inputs, and most of all help protect life and minimize or eliminate physical injury? Well think about the software efforts NASA engineers have to define, create, test, and deploy with ongoing refinements in the Orion Deep Space vehicle that will carry astronauts to Mars and back controlled by one of the most sophisticated software endeavors I have ever seen.
We are very fortunate to have Darrel G. Raines, NASA Orion Flight Software Production Systems Manager as the keynote speaker at the Embedded Systems Conference (ESC) Minneapolis on November 4 this year. Raines’ topic will be “Orion Multipurpose Crew Vehicle Overview—Embedded Flight Software”
I posed six question to Raines which I feel are of significant importance to the Orion avionics systems as well as to our audience.
Question #1: Re-use of existing components from Boeing’s 787 aircraft is a key factor that will speed up Orion development with a tested computer system and “Glass Cockpit” display system. How close in design are Orion’s devices to the actual devices used on the 787? What are the differences that Orion needed for operation in space?
Raines: The flight computers that are used on Orion are direct descendants of the ones used on the 787. The fact that Honeywell has such a good product with that flight computer seems to have weighed heavily in the favor of Lockheed Martin with the Constellation selection committee. However, there are some major differences between the two designs. The Orion version has to handle greater physical stress (launch), larger and more frequent doses of radiation, and longer periods of use prior to replacement or repair (fault tolerance). The current design was modified from the original to achieve those goals.
Question #2: What is the advantage of Orion’s Centralized Command/Control System architecture vs. previous spacecraft Distributed architectures such as the Space Shuttle had?
Raines: The Space Station has a very distributed computer architecture. That is necessary so that local computers can take on local tasks. The station also benefits from this design when you consider potential catastrophic failures. A fire or other damage in any one section will not completely shut down the station computing facilities. The Orion flight computer design is built upon the principle of a set of centralized computers with all command authority. However, the system has built-in fault handling within each computer (a self-checking pair design) and by having those computers fail-silent. Such a design emphasizes transaction speed and throughput.
Question #3: Orion uses a relatively slow processor as the basis for onboard computing. Why is this so and does it cause problems in getting done all the computation necessary for a spacecraft?
Raines: The fact of life for spacecraft design is that we never use state-of-the-art processors. Certainly when a design is begun, we use current generation designs. But a processor that will be used in space has to be certified to perform with all of the constraints I mentioned in the first question: vibration, radiation, fault tolerance, etc. Therefore, we typically look for designs that are more reliable than they are fast. Furthermore, the typical design/production/flight/maintenance life cycle of a spacecraft can be as long as 20 to 40 years. Think of how slow you view a laptop that is 5 years old. Orion has been in development for 9 years. So, yes, the computer is relatively slow. But it will seem a lot slower in 30 years when we are still using the spacecraft.