Early Mars rovers had little more intelligence than a fancy remote-controlled car. NASA’s Curiosity rover is somewhat more evolved: It can navigate around simple obstacles and spot a dust devil on its own.

Much more brainpower would be required for a robotic exploration of Saturn’s moon Titan. Home to one of the solar system’s liveliest environments outside of Earth, Titan has tidal seas of methane, a stormy atmosphere and perhaps ice volcanoes. In 2005, the European Space Agency’s Huygens probe landed on a beach in Titan and transmitted data about its surroundings for 90 minutes; ever since, scientists have been eager to send another probe with more staying power. But because communications from Earth to Titan take hours, the robot would have to solve problems on its own. If it has been floating in a methane sea for months and finally approaches shore, it must begin taking pictures of newly visible land. If a methane-dwelling octopus swims by, engineer Trey Smith of NASA’s Intelligent Robotics Group says, it must notice. He’s only mostly joking.

With the aim of building a robot smart enough for Titan, Smith and a team of other engineers and scientists spent three weeks this month at a remote lake in the high Chilean Andes. They were field-testing the Planetary Lake Lander, an early prototype of a floating space probe that could, among other things, notice an octopus swimming past.

Since no hydrocarbon seas are available on Earth, the researchers, from NASA Ames Research Center and the Search for Extraterrestrial Intelligence (SETI) Institute in Mountain View, Calif., chose to test their probe on a lake under the disappearing Echaurren glacier, an ideal setting for technology that must notice both abrupt and subtle environmental changes.

The robot has been exploring the lake with increasing astuteness since 2011, with periodic visits and upgrades from its makers. In August the team endowed the robot with the ability to collect and communicate meteorological data. Since then the team has received six emails from the Lake Lander, each accurately announcing a storm on the lake. In one case, the accompanying photograph sent by the robot showed its camera buried in snow.

“My colleague posted on Facebook that he had just sent Curiosity its instructions for the day and I wrote, ‘My robot is smarter than yours—he’s the one sending me emails!’” says Principal Investigator Nathalie Cabrol of the SETI Institute. “We will stop it short of writing our papers.”

The Lake Lander has a long way to go before it can write scientific studies, but it is already mastering some adaptive science techniques. It has learned to notice irregularities in its environment and focus its instruments on those anomalies. It conducts simple data analysis and sends only the most interesting information to scientists.

Selectivity in transmitting data is crucial for a robot on Titan. It takes about a joule of energy to send one bit of information from Titan to Earth, says NASA roboticist Liam Pedersen. At that rate, sending a single black-and-white compressed image would require the energy equivalent of a D-cell battery. A probe on Titan must therefore be stingy about what it sends and must compress information intelligently based on what it has learned about its environment, Smith says.

These technologies have applications on Earth as well. Selective data capturing and transmission could be used, for example, by probes that monitor ocean chemistry or by telescopes that scan the skies and send alerts when they spot pulsars, quasars, supernova or asteroids.

“I don’t know if we are smarter than the dinosaurs, but we have more technology than they did,” Cabrol says. “If we want to continue surviving on this planet, we’re going to have to get better at monitoring our environment.”

As it lays the groundwork for a future Titan mission, the Planetary Lake Lander’s technology is also monitoring the rapid decline of Echaurren glacier, a victim of climate change whose remnants frost a toothy ridge 1,000 feet above the lake. For the team’s recent visit, burros carried in equipment from a nearby dirt road. Thunder frequently emanated from a cloud-clotted volcano nearby, momentarily drowning out the assemblage of small generators that buzzed the camp to life. Otherwise the only visible mark of humankind was the canary yellow robot reflected in the cobalt lake.

In November the NASA Astrobiology Science and Technology for Exploring Planets program notified Cabrol’s team that the Lake Lander project had been extended for another year. The scientists plan to use the time to advance the adaptive science system. Yet whatever progress they make, the window will soon close for a robotic visit to Titan’s largest seas.

In 2012 a proposal to drop a probe into Titan’s Ligeia Mare narrowly lost a NASA funding bid to a 2016 Mars lander mission. The Titan Mare Explorer is still seeking funding, but if it does not come through this year, it will be too late—the robot must launch in 2016 if it wishes to arrive on Titan before the polar area loses direct communications with Earth in 2025. It will not regain contact until 2040.

Waiting 15 more years to explore Titan’s seas would be a shame, since they could host life, Cabrol says. She concedes that life may not exist in a methane-drenched world, but you never know. “Methane has its problems for life, but that being said, it could just be that our intellect is limited,” she says. “And if I had to bet, I would bet on that.”