In the end, the mission was heralded as a success — the first American moon landing since the Apollo 17 mission in December 1972 and the first by a commercial venture.

But the descent to the surface was a perilous nail-biter that required on-the-fly ingenuity to save it from failure. And it serves as a reminder that space travel remains an extraordinarily risky endeavor and that the moon, even 50 years after the U.S. first landed astronauts there, remains a forbidding and elusive target.

On Friday, Intuitive Machines, the Houston-based company that designed and operates the lander, said the vehicle was “alive and well” and that “flight controllers are communicating and commanding the vehicle to download science data. The lander has good telemetry and solar charging.”

But that was not certain on Thursday afternoon as the spacecraft was to begin its descent. That’s when ground controllers realized they had a big problem: the sensors on board their Nova-C spacecraft, dubbed Odysseus, were not working. Without them, the spacecraft was essentially flying blind, unable to navigate the hilly and rocky terrain below. If Odysseus were to touch down softly on the moon, Intuitive Machines would have to fix this — and fast.

Ground controllers, searching for a way to fix the problem, needed more time. So they commanded the 14-foot-tall spacecraft, which looks like a phone booth on stilts, to orbit the moon once more. Intuitive Machines announced that the landing, which had been moved up to 4:24 p.m. when everything appeared to be going well, was being pushed back to 6:24. It didn’t say why.

The extra orbit, ground controllers hoped, would buy them the time they needed.

NASA had known from the beginning that its decision to send a fleet of privately developed robotic spacecraft to the moon was a huge risk.

The program, called the Commercial Lunar Payload Services program (CLPS), was unlike any deep-space program NASA had ever done. NASA wouldn’t own or operate the spacecraft. NASA would be merely a paying customer, hiring a fleet of robotic space taxis developed by an array of commercial companies to transport its instruments to the moon. The hope was that along the way, the companies would develop the technologies that would allow the space agency — and its industry partners — to visit the moon more frequently and for a lower cost. If successful, the program would help pave the way for landing astronauts there as part of NASA’s Artemis program.

Intuitive Machines was working under a $118 million contract from NASA under the program. Last month, the agency’s first CLPS mission, carried out by a Pittsburgh company known as Astrobotic, suffered a propulsion problem, lost fuel and did not reach the moon. Now, it seemed Intuitive Machines might not make it either.

The sensors that Odysseus intended to use to find a landing site were out. But ground controllers knew they had a backup: a 33-pound instrument developed by NASA affixed to the outside of the spacecraft like a large barnacle on the hull of a ship. The system was not supposed to be used to guide the spacecraft to landing; rather it was on board as an experiment to see if it would work and then could be used for future landings.

Called the Navigation Doppler lidar (NDL) for Precise Velocity and Range Sensing it would use lasers to calculate velocity and the distance to the ground during the descent. It worked like the radar detector police use to nab speeders, but it would use “pulses of light from a laser instead of radio waves and with very high accuracy,” Intuitive Machines had explained before the mission.

The instrument was developed by a team at NASA led by Glenn Hines at the agency’s Langley Research Center in Virginia. Hines said in an explainer video posted online that the NDL instrument has “three telescopes where light would come out of the telescope, hit the moon’s surface and then some of that light will be reflected back.”

In the Apollo era, he explained in the video, NASA had relied on “large radars” or even “astronauts using their eyes,” such as when Neil Armstrong famously took control of his spacecraft during the Apollo 11 moon landing in 1969 and guided it to a safe spot as fuel nearly ran out.

The NDL instrument “is going to help take the burden off the crew with a much smaller, lower power and more accurate instrument,” he said.

While Odysseus remained in lunar orbit, ground controllers scrambled to develop a software patch that they could then shoot up to the vehicle directing it to feed the readings coming from the NDL system into the spacecraft’s computer, instead of using the primary navigation system.

With the broadcast of the mission now live on NASA’s website, Josh Marshall, Intuitive Machines’ director of communications, explained that the company had been working a “dynamic situation,” swapping out the sensors on the fly, with the spacecraft some 240,000 miles from earth.

“We had to improvise a little bit,” he said. “And it sounds like we are getting good readings from those images — an absolutely remarkable feat.”

The crews “working to patch that software were certainly under pressure,” he added. “The clock was ticking as we went into that extra lunar orbit. It wasn’t a situation where we could just sit in lunar orbit and try to solve our problems indefinitely.”

The landing was tense. The spacecraft fired its engine for 11 minutes, throttling back power as the vehicle burned through fuel and got lighter and lighter. The landing time, 6:24 p.m. came and went without confirmation of a landing. It was unclear whether the spacecraft had survived, or what condition it was in.

“We are in standby mode,” Marshall said.

“Checking antennae reception,” he reported shortly thereafter.

About two minutes after the intended landing time, Mission Director Tim Crain called out to his team: “All stations, this is M.D. Please look back through your logs and confirm the last information you had, and we’ll determine if this is a comms outage.”

A few minutes later, Crain polled his team again. “Looks like we had excellent pitch and yaw control throughout,” he said, referring to the orientation of the spacecraft during the descent. “But it did see a little bit of a roll excursion. Could it be that we landed off-angle in the final phase?”

Maybe the spacecraft had landed and tumbled? As the teams tried to find out, they also scrambled to reestablish communications.

Finally, about 10 minutes after the intended landing time, Crain had good news: “Signs of life,” he said. “We have a signal we’re tracking.”

The signal, he said, was “faint, but it’s there.”

A minute later: “What we can confirm without a doubt is our equipment is on the surface of the moon. And we are transmitting. So congratulations, IM team.”

The room broke into applause.