Chapter 288: Isn’t Your Progress A Bit Too Fast?
Cheers erupted in the Wenchang Control Center. Everyone was a professional and clearly understood the difficulty of this soft landing.
Simply put, the Americans conducted 11 manned moon landings, with 6 successfully delivering astronauts to the lunar surface, but landing on the edge of the Lunar South Pole was a first of its kind.
Far away in the Apollo Technology meeting room in Kunshan, it was already early morning Yanjing time, and the Russian experts at the tables all had coffee; without coffee, they really couldn’t hold on.
They were still in Kunshan at this time and hadn’t moved to Shanghai yet, because on one hand, the office building there hadn’t been vacated, and on the other hand, the security work wasn’t fully completed.
As the commotion caused by Apollo Technology grew bigger, the entire security level was constantly upgrading. At this current time point, it was about ensuring absolute safety within a 20-kilometer radius around the office area.
So Apollo Technology was expected to move there by the end of this year.
Kunshan and Gusu were extremely envious; their hard-earned unicorn had been snatched away by Shanghai.
Back at Apollo Technology, the Russian experts looked at the data and video footage transmitted from Wenchang, and one by one they were dumbfounded.
They knew Apollo Technology was impressive, but they didn’t expect it to be this perverted.
To put it this way, in general understanding, your soft landing at the Lunar South Pole this time was a first of its kind, so it would definitely prioritize stability.
Simply put, it would definitely use automatic navigation, with remote intervention only in case of accidents, and then gradually accumulate experience with more launches to optimize the automatic navigation proposal.
This is normal logic, but Apollo Technology’s proposal is not normal.
Although they also have remote intervention, this remote intervention isn’t about ensuring Wu Gang 0001 completes the soft landing smoothly, but just ensuring it lands at the Lunar South Pole; whether it’s a soft landing or not doesn’t matter.
Either fully automatic navigation achieves a soft landing, or rely on my remote intervention for landing.
As long as the landing position is the edge of Lunar South Pole Shackleton Crater.
In the end, they really watched with their own eyes as Wu Gang 0001 completed the soft landing throughout, performing one high-difficulty movement after another, and smoothly landing on the edge of Shackleton.
“No, this is of course extremely difficult. The entire process requires precision in navigation and guidance, from launch to trans-lunar injection to mid-course correction to lunar orbit insertion, all demanding very high precision.
Although China’s Long March series achieved precise launches, Earth control center intervention is still needed during the mid-course.
And in terms of mid-course correction capability, China’s Chang’e series demonstrated mid-course correction capability. If fully automatic, it would require even stronger robustness.”
Alyosha and Alexander were both experts who came along this time.
The former was responsible for orbital calculation during rocket launch, the latter participated in communication technology.
After glancing at each other, the two began discussing in low voices.
After Alyosha finished speaking, Alexander added: “I’m not saying these two stages aren’t difficult, but that the subsequent stages are even more difficult.
Whether Earth orbit insertion, trans-lunar injection, mid-course orbital correction, or lunar orbit insertion, they all require precise calculation.
In the past, China Aerospace’s precise calculation capability scored 80 points; now Professor Lin and they merely raised that 80 to 90 or even 95.
This already has a foundation; from a technical perspective, there are many solutions, just theoretical solutions and their applicability in reality.
Professor Lin, as a mathematics master and aerospace expert, his judgment is beyond doubt; he just selects and can choose the best proposal to optimize past technology.
I think the biggest difficulty is still in descent and landing.
This process has no experience to reference; neither China Aerospace nor NASA has it.
The South Pole region is full of high mountains and craters, with permanently shadowed regions in constant darkness. If you use a vision navigation proposal, the low-angle sunlight at the edges will have a serious impact.
Think about it: the temperature in lunar shadow regions is -203 degrees Celsius, while the sunlit areas are 54 degrees. It’s hard to find similar scenarios on Earth for testing.
This is the hardest part.
As for NASA’s Lunar Node-1 proposal, it’s only at the theoretical level; in reality, putting it in such a complex scenario would make it completely unusable!” Alexander shook his head, his face full of shock and admiration.
Everyone used to score around 70; at most, in recent years with money, resources, and investment, China jumped from 70 to 75—this is about overall aerospace—and then quietly, the opponent produced a pervert who scores 95, far ahead of the previous first place, NASA’s 80.
The Russian experts couldn’t help but be shocked.
The Lunar Node-1 proposal mentioned by Alexander was proposed by NASA, relying on radio signals to support precise geolocation among landers, ground infrastructure, and astronauts, providing navigation observation services in digital form to ensure they can quickly determine positions relative to other spacecraft, ground stations, or moving rovers on the Moon.
This proposal is mainly used in space to assist lunar spacecraft with orbital maneuvers and guide landers to successful landings on the lunar surface.
(The image shows a lunar landing device equipped with Lunar Node-1 signal sensors)
But the prerequisite is that you need enough signal transmitting and receiving units on the Moon, mutually assisting to complete the construction of this system.
This is also part of the series of lunar navigation infrastructure that America plans to build on the Moon.
“Imagine getting verification from the lighthouse on the shore you’re approaching, rather than waiting for news from the home port you left days ago,” said Evan Anzalone, navigation system engineer at NASA’s Marshall Space Flight Center in Huntsville, Alabama, and chief researcher of the technical proposal, in an interview: “What we seek to provide is a lunar network of lighthouses, offering sustainable local navigation capabilities, enabling lunar spacecraft and ground personnel to quickly and accurately confirm their positions, rather than relying on Earth’s control center.”
Of course, it’s still on Earth and hasn’t gone to the Moon yet.
If Lin Ran were still working at NASA, using the gate, then building small sensors and directly dropping the sensors up there, the system would be initially set up; no need for such trouble.
As for NASA’s system, first it’s only on Earth, second they need to be able to shoot things to the South Pole edge; they haven’t even done the first step, far from success.
That’s why the Russian experts think this thing is just talk on paper.
And what they’re seeing now is Apollo Technology’s automatic navigation directly achieving the hardest South Pole edge soft landing.
Everyone wants to know how exactly you did it.
Valentin was no exception. Sensing the whispers among the experts he brought and their inner longing, he asked: “Professor, this is truly a remarkable achievement. Apollo Technology has created another miracle. Please allow me to extend my sincere congratulations.”
Valentin’s flattery was very sincere, both because he was genuinely convinced after watching the entire process, and because Apollo Technology’s achievement was beyond doubt.
Scoring Top 2 in the college entrance exam, others praise you with boundless future; scoring for junior college, others praise you with boundless future—even if both are sincere compliments, the latter sounds like sarcasm.
“But Professor, could you explain to us how you did it?” Valentin asked. “We are all very curious.”
Lin Ran thought for a moment, then said: “Regarding this, we used too many technological innovations.
I’ll just pick a few points that I think everyone will be interested in and talk about them.
I’ll mainly talk about the innovations we made in the algorithm field to improve overall navigation precision.
We used convolutional neural networks for lunar terrain relative navigation to perform crater detection at the vision level.
Terrain relative navigation can improve spacecraft position estimation precision by detecting global features, which act as supplemental measurements to correct drift in the inertial navigation system.
We mainly used convolutional neural networks and image processing methods to build a set of algorithms that track the simulated spacecraft position through an extended Kalman filter.
This allows intuitive detection of craters in simulated camera frames during the process and matches these detection results with known lunar craters in the region of the current estimated spacecraft position.
These matched craters are regarded as features tracked using convolutional neural networks.
Thus, this system enables more reliable position tracking for image brightness changes and more repeatable crater detection frame by frame throughout the trajectory.
When tested on trajectories using standard brightness images, compared to Kalman filters using image-processing-based crater detection methods, the new method reduced average final position estimation error by 90% and average final velocity estimation error by 50%.
Oh, by the way, you can see this method in a paper accepted at the 2020 American Control Conference; we made some small optimizations in that paper.
Through this algorithm, we ensured we could detect craters and rocks and find flat ground.”
The quick Russian experts had already started searching on their notebook computers.
“At the sensor detection level, we cooperated with technology companies in our country; they have rich experience. We combined lidar, camera, and IMU data, using particle filter and Kalman filter algorithms to fuse multi-source data and reduce single-sensor errors.
Alright, I’ll keep it simple: this is mainly a lunar lander navigation solution based on Terrain Relative Navigation method.
Algorithms were developed on a scaled simulated lunar scenario, with a three-axis motion framework built in that background to reproduce the landing trajectory.
At the tip of the three-axis motion frame, long-range and short-range infrared ranging sensors were installed to measure height.
We all know that calibration of distance sensors is crucial for good measurement results.
To this end, sensors were calibrated by optimizing nonlinear transfer functions and bias functions using the least squares method.
Therefore, the sensor covariance is approximated using a second-order function of distance.
These two sensors have two different operating ranges, overlapping in a small region.
To achieve optimal performance in the overlap range, a switch strategy was developed.
After evaluating the switch strategy, a single error model function for distance was found.
Due to different environmental factors, temperature deviation at crater edges is large, so bias drift of the two sensors is evaluated and appropriately considered in the algorithm.
To reflect lunar surface information in the navigation algorithm, a digital elevation model of the simulated lunar surface has been considered.
The navigation algorithm is designed as an extended Kalman filter that uses height measurements, digital elevation model, and acceleration measurements from the motion coordinate system.
The goal of the navigation algorithm is to estimate the position of the simulated spacecraft during landing from 3 km height to the landing point near the crater edge.
And continuously update the algorithm during landing; for this, we specifically designed a crater peak detector to reset the navigation filter using new state vectors and new state covariances.”
Everyone listened very attentively.
At this time, Alyosha had found the American Control Conference paper mentioned earlier by Lin Ran. Alexander glanced at the abstract and muttered: “Pervert!”
Alyosha didn’t ask why pervert.
Because the abstract in the American paper stated an average final position estimation error reduction of 60% and average final velocity estimation error reduction of 25%; in Lin Ran’s version, the so-called small optimization resulted in 90% error reduction.
The two Russian experts racked their brains but couldn’t figure out how China achieved this small optimization.
“Regarding landing precision, everyone knows that for our launches, we ultimately need to achieve spacing between fuel tank and lunar module positions not exceeding 200 meters.
Including this landing, I believe everyone saw that the error between our target point and actual point should not exceed 20 meters.
Our limit can even be lower than 20 meters.
Each landing is at adjacent positions to ensure moon base construction can use existing resources as much as possible, with every spacecraft launched to the Moon put to use.
This is also built on the shoulders of predecessors.
This proposal should initially be attributed to Capuano’s work in 2015; they studied code-level Earth navigation system signal receivers for precision assurance during landing across the entire lunar orbit, achieving 700 meters precision in that study.
That is, using Earth navigation system signals to support lunar missions—you’ve probably heard of it, after all, the European Space Agency researched GNSS receivers in 2021 for use on the ESA-SSTL Lunar Pathfinder spacecraft, reducing precision to 100 meters.
At that time, you hadn’t fallen out with Europe yet; many of their projects would have been shared with you.”
GNSS: Global Navigation Satellite System, that is, global navigation system; GPS, Russia’s GLONASS, Europe’s Galileo, and China’s Beidou all fall under this category.
The Russian experts present were a bit embarrassed.
What does that mean? Have we fallen out now? We just temporarily stopped cooperation; cooperation will resume soon.
These Russian experts, deep down, still hoped to integrate with Europe; on one hand, it’s the Russian nature, on the other, they had extensive cooperation with European peers in past work—who doesn’t have a few European expert friends?
Lin Ran ignored their expressions and continued: “Later, there were successive developments for guiding lunar probes to the Moon requiring real-time accurate position and velocity information, especially in approach and braking phases, with navigation info provided by ground tracking stations, including S-band ranging, Doppler systems, and very long baseline interferometry. Our China experts in the Chang’e Five series proposed an intelligent heterogeneous sensor data fusion method for lunar probe descent navigation, achieving kilometers-level positioning precision.
On the wisdom of predecessors, Apollo Technology built a proposal for autonomous navigation of lunar spacecraft landing using lunar gravity gradient strategy measurements in the approach phase.
As the spacecraft approaches the Moon, the gravity gradient signal strength increases.
The gravity gradiometer onboard the spacecraft can precisely measure local gravity gradients and use the latest lunar gravity model for reference values.
Considering the decrease in spacecraft altitude, the truncation degree and order of the gravity model are gradually increased to balance computation cost and model precision.
We developed an iterative Kalman filter for orbit and attitude coupled estimation using gravity gradient measurements and attitude quaternions from star sensors.
While considering gradiometer noise levels.
We conducted simulation tests before this launch, and the results showed that the spacecraft position converged rapidly, reaching less than 10 m precision in the final period, i.e., during landing.”
The room erupted in an uproar.
The precision mentioned by Lin Ran—the previous 100-meter precision—was orbital precision.
While the final landing precision, as Lin Ran also mentioned, in China Aerospace’s proposal was kilometers-level orbital precision.
Wait, orbital precision of kilometers, how did it become 10 meters for you?
From kilometers to 10 meters—isn’t that span a bit too big?
Everyone couldn’t understand.
What’s even more exaggerated is that your simulation result of 10 meters is one thing.
After all, simulation is simulation, actual landing is actual landing.
But your actual landing effect now also achieved within 100 meters.
How exactly did you do it?
Moreover, Lin Ran explained in great detail; for outsiders, this level is absolutely sufficient.
But not so detailed that even knowing this much, they felt they couldn’t replicate it.
Valentin knew he shouldn’t ask—from no angle should he probe others’ secrets—but he couldn’t hold back: “Professor, could you tell us about the specific algorithm design?”
Lin Ran’s expression changed: “Of course not!
I’ve already told you our technology evolution path, the thinking process, which previous papers we used, and the core ideas of our algorithm design like gravity gradient.
Asking further is impolite!”
Valentin immediately apologized: “Sorry, I was presumptuous, Professor. Please forgive my presumption, because what you achieved is so, so unbelievable.
We’ve never seen such a huge technological leap.”
