![]() Although ChatGPT might not meet some special elevated standards for what is considered general AI, it is able to solve a wide variety of generalized requests, including writing this response on command. Here's another recent HN reference point on capabilities that don't contribute to a great product: The person who (with the PI, Bruce Banerdt) led the InSight "requirements trace" is Peg Frerking ( ), who is now a JPL fellow in part because of her work in adding engineering rigor to the formulation process.Īs this thread shows, it's very easy for clever engineers to add on nice-to-have's that are not required to meet science objectives. The whole presentation is great: the first introductory slides, some InSight examples, and the last 3 slides in particular give background on some of my claims above. Here's a description of how this works in general, from someone who was involved in InSight formulation and who is in the leadership of mission formulation at JPL: For the seismograph, I think they had an approximate event rate and they needed to capture X number of events of a given magnitude to get insight (heh!) into the subsurface structure. In the case of the heat probe, they needed to observe at least one full annual heating cycle. InSight didn't need more than one Mars year for the science, so that's that. The science objectives are carefully time-bound by a very knowledgeable science team, and from those science requirements, all the engineering requirements, including lifetimes, are derived. You have to come in under a cost cap, while demonstrating you will achieve science objectives. Within competed programs, you can't spend money on nice-to-have's. To add context: InSight was proposed to a cost-capped, competitive NASA program. ^- This is the answer: this capability wasn't needed to fulfill science objectives. Now that we have superhuman levels of general AI in chatgpt, isn't it time to work on foolproof and lightweight robot arms for all sorts of tasks? If a human could reach through a portal to there with a brush, it would take less than five minutes to brush off the dirt that had fallen on solar cells.Ī simple, light, and foolproof robotic arm that can easily do that doesn't exist. Robotics isn't at the level where that is easy, cheap, or lightweight. ![]() The lander weighs 789 lb and the cost-effective solution is not try include a simple cheap lightweight robot arm that is foolproof and can easily sweep some dirt, because there is not such thing. So robotics is at the state where it is possible to send a rover to Mars, but a simple robotic arm can't easily brush off some dirt. The simplest, most cost-effective way to meet my goals was to bring solar panels big enough to power my whole mission – which they did (and then some!)." I went out of it thinking, "yep, look at these excuses we are making for the state of robotics." - the quote is: "A system like that would have added cost, mass, and complexity. prove me wrong." I mean how hard could it be? No way is that an "impossible" task. Planned obsolescence is perhaps a bit too strong of an accusation though, so I'm sorry for saying that.Īfter reading your note "why it's not possible for the solar panels to be self cleaned" I clicked through and as I clicked I was thinking "okay so obviously it is possible for the solar panels to self-clean. It just sets off my code smell detector too much. So I'm afraid that there's still no way I can accept that the engineering isn't feasible. Kind of like the old joke about NASA spending millions to design a space pen when the Soviets just used a pencil. It may not be glamorous or even guaranteed to be environmentally friendly, but it would work. That said, I also realized that a motorcycle helmet tear off plastic film and some rubber bands held stretched by fusible links would solve the problem just fine and at least double the life of the mission for negligible weight. ![]() ![]() Who are we armchair warriors to dismiss that process? Probably every engineering team at NASA who was part of Insight had a say in whether to invest an ounce in a dust cleaner or another sensor, and they went with the sensor. Hey nradov after sleeping on it, I agree with you, I don't like the tone of my original comment.
0 Comments
Leave a Reply. |