View My Stats

Wednesday, February 08, 2006

Both Robots and Humans needed on Mars

Originally written on May 10, 2004
Updated: Feb 8, 2006

A couple of points before we delve into the details -
  • Point 1: It's not about whether Robots or Humans should explore Mars - Robots and humans will have to work together on Mars to have a productive mission, doing what each one of them are best at doing. Humans with their immense creativity and ability to improvise on-the-spot when things fail. Robots, in their ability to do boring stuff accurately again and again without getting bored or making mistakes, or forgetting (assuming your robot software does not have too many bugs !).
  • Point 2: Often many people don't realize that you cannot do tele-robotics (remote controlled robots) on Mars from Earth. But you CAN do tele-robotics with robots on Mars, if the human operator is also on Mars. Light takes anywhere between a few minutes to half-an-hour for one interactive command from Earth to Mars! This is the reason the robots Spirit and Opportunity sitting on Mars are so slow - their decision makers and controllers are sitting on Earth. And it takes days for the Humans to plan a few seconds of work on Mars, since it cannot be interactive, as opposed to operating a crane or a dump-truck on Earth. Any amount of AI cannot remove the human decision from the loop besides reducing it, unless the robot is as smart as humans. Geologists and biologists need to hold rocks and fossils in their hand to be sure.
  • Point 3: Given 1 and 2, the cost of per unit exploration by robots is significantly more expensive than humans, because of their (lack of) speed, crudeness in mechanical dexterity, and intelligence smaller than a cockroach. The fact that you have team of 100s of people (400 to be precise for the current mission) support 2 tiny robots on Mars just tells you that it will be hard to scale the exploration up. Any improvement in the speed/efficiency/cost of robots also makes the human missions cheaper by the same amount, since more automation will allow people on Mars to spend more time on exploration than maintaining the robots and machinery. The ratio of the cost of the two mission converges not diverges - the human mission costs converge to the robotic mission costs as robots become infinitely as advanced as humans - in such a situation you could let the robots do everything on Earth - design the spaceships, build them and launch them - no PhDs, scientists, engineers or human workers needed! In fact no humans needed! And once you train a single robot to do rocket science, you can make a million copies of it. The additional cost soon approaches zero. Sending a few humans along is practically free, since the food and material consumed by them is being produced by other robots. Contrary to this super-robotic scenario NASA currently spends most of its money paying salary to people to develop the hardware and software (and so does all the other companies in the world)



Here is an email I wrote in response to a well-known "anti-manned-space-exploration" personality. One cool thing about having humans on Mars is that you don't need to have an army of 400 engineers operating one robot on Mars remotely. According to the latest plans, having humans on Mars, 21st century automation (unlike 60s and 70s technology in the Apollo and Space Shuttle), and the minutes of communication lag will lead to a model of "Mission Support" not a "Mission Control" thus requiring a relatively small number of engineers on Earth.

This article also contains some rough cost comparisons I did for humans vs. robots.

Current Human exploration cost: 3,25 million dollars per man-day of exploration. Current Robot exploration cost A (200 times slower than humans): 400 million dollars per man-day !!

Current Robot exploration cost B (10,000 times slower than humans): 20 billion dollars per man-day = 20 billion for one days work ! Wow, even Bill Gates will go bankrupt paying someone like that for 3 days! Note: if Spirit and Oppurtunity survive 4 years with declining productivity these numbers will only change by a factor of 3 or so, still making them super-expensive compared to humans.

Finally, any manned exploration will extensively use robotics and automation to explore Mars enough to answer the big questions. Without humans, you get only "teaser" exploration and high costs per unit work. Humans and robots working together on Mars is the key - as it lowers the cost while still giving us the flexibility and speed of the human brain. Robots can also help prepare a site for human missions without having the humans worry about the preparation. Read Mars Direct and NASA's reference plan (derived from Mars Direct) for more info on how this will work.