This interesting story about the Mars Curiosity Rover recently came to my attention – that its software has now been upgraded to allow the on-board computer to make decisions about choosing targets for its laser to zap. You can read the NASA press release by following this link.
It interested me for two reasons. Firstly, it is a reminder that we have two rovers actively studying Mars as I type (the Mars Curiosity Rover and the Opportunity Rover, which has been operating on the surface of Mars since January 2004!). Whereas Opportunity is about the size of a shopping trolley, the Curiosity Rover is about the size of a car, and has a whole suite of scientific instruments to learn more about the geology of Mars. This includes a chemical laboratory, which can analyse the composition of rocks. The laser is used to vaporise nearby rocks which are thought to be of interest. As the laser strikes the rock the gases emitted are analysed by a spectrometer, but Curiosity can also scoop up rock samples and place them in an on-board oven to heat them up and further analyse them.
But, the second interesting thing for me is that this marks a step forward in “artificial intelligence”. Now, I am very far from being an expert in “artificial intelligence”, so someone who knows more about it than I do may well correct much of what I am about to say. However, it is clear that the on-board computers on the Mars Curiosity Rover are now making decisions about potential targets for the rover’s laser, presumably based on analysing images of previous rocks which Mission Control (at the Jet Propulsion Laboratory in Pasadena) have chosen as targets. Thus, the computer has been learning which kind of rocks the geologists/experts on Earth have been choosing, and is now choosing its own based on some criteria of similarities. I find this very interesting (maybe I’m easily pleased!)
Such computer-based learning and decision making are vital as we continue to explore the Solar System with robotic missions. The delay time between sending a command to a robot on a distant world and getting the response becomes longer and longer as we explore more distant planets and moons. With Mars the delay is not too bad, typically 20 minutes, but with Saturn it is more like 160 minutes between sending a command and getting the response. Nearly 3 hours is too long in some circumstances, so a rover on e.g. Titan in the future would need to be able to make some decisions on its own, after a period of being commanded and learning from those commands.
Having rovers which use artificial intelligence is, in my opinion, still no substitute for having a human being on Mars. The work which the various rovers on Mars have done in the last 12 years could have been accomplished by a skilled geologist in a few days. And, as the excitement over Tim Peake’s 6-month spell on the International Space Station has reminded us, nothing gets us more excited in the matter of space exploration than seeing a human being doing things in space; no matter how impressive are the things that robots can now do.Hopefully, I will see human beings on Mars in my lifetime.