Europe's Mars lander failed due to hardware and software inadequacies, according to a new report:
The crashed European spacecraft Schiaparelli was ill-prepared for its attempt at landing on the surface of Mars. That's the conclusion of an inquiry into the failure on 16 October 2016.
The report outlines failings during the development process and makes several recommendations ahead of an attempt to land a rover on Mars in 2020. That mission will require more testing, improvements to software and more outside oversight of design choices.
[...] The report authors catalogue a series of necessary upgrades to onboard software, as well as suggesting improvements to the modelling of parachute dynamics. They also recommended a more stringent approach - including better quality control - during the procurement of equipment from suppliers.
One of the recommendations is that NASA's JPL should double-check ESA's work.
(Score: 1) by khallow on Friday May 26 2017, @03:14PM (6 children)
(Score: 2) by kaszz on Saturday May 27 2017, @06:09AM (5 children)
That would fail on cost. And the wasted travel time.
It's far cheaper to spend some more thinking on simulation and data analysis. Plus proper specification and delivery control with subcontractors.
(Score: 1) by khallow on Saturday May 27 2017, @10:44AM (4 children)
Making more than one copy right away increases the value of the effort put into development of new spacecraft - a mostly one-time cost per vehicle design. Plus, it allows for typical economies of scale from making more than one or two items at a time.
Second, the travel time is wasted whether or not the spacecraft is built. A common example I like to give is the development of the Mars Exploration Rovers [wikipedia.org] followed by the Mars Science Laboratory [wikipedia.org]. The two MER were sent in 2003 followed nine years later by the MSL in 2012. However, NASA could have sent for the cost of the MSL, about eight [slashdot.org] MER instead and even with the cautious schedule I mentioned before (sending a couple of spacecraft every two years when a launch window becomes available) have them all on the surface of Mars (well, aside from the failures, of course) gathering data by the time that the MSL was supposed to lift off from Earth.
Notice we are once again on the same extremely long time track with the MSL successor mission, currently named Mars 2020 [wikipedia.org] which isn't expected to launch until mid 2020. That mission could have launched in 2012 or 2014, if NASA had made more than one MSL in the first place (the new rover uses the same design and some of the spare parts).
So fail on cost? That depends on what you think the point is of spending money on spacecraft and such. If it is merely technology development for dead end missions that will rarely be replicated, then the current approach may well be better. Wasted travel time? We lose the better part of a decade on development of the next generation.
If you're interested in space exploration and science, batch manufacture is both better science for the money spent and more resulting output over the same period of time than the current model of one or two per generation of mission.
(Score: 2) by kaszz on Saturday May 27 2017, @01:59PM (3 children)
My point is that trial and crash is a very inefficient method to do development feedback for space exploration. But obviously there already is something that works so maybe they should continue using that. While in parallel developing better capabilities. Because if I recall correctly the current generation has weight limitations. But that will probably require a rocket lander ie like SpaceX uses.
(Score: 1) by khallow on Sunday May 28 2017, @02:26AM (2 children)
Compared to what? For example, your proposed alternative, simulation and data analysis, has the weakness that those methods are unreliable without real world feedback - which is trial and crash, ultimately. And the various space agencies' collective tendencies are to construct technologically advanced vehicles which by their nature and novel environmental challenges are the hardest to successfully model in these ways.
My view is that it is mere luck that NASA, for example, hasn't recently lost a spacecraft to error or malfunction. Each time that failure happens, scientific progress in that particular venue is stalled by five to ten years in the present approach (while it would merely be delayed till the next viable launch date for my approach, perhaps even sooner, if ongoing missions can merely be reprogrammed to avoid the failure), and if NASA chooses not to replace the mission, that delay can go up to many decades or more.
(Score: 2) by kaszz on Sunday May 28 2017, @05:30AM (1 child)
It's not a binary choice. But to do trial and error when there are simulation tools that can advance sufficiently. It makes more sense to use them first fully. And then test.
(Score: 1) by khallow on Sunday May 28 2017, @12:19PM
Looks like they weren't advanced enough in this case. Simulation tools cost money too. I don't buy that they're automatically cheaper than just making more stuff and using it.