NASA Autonomous Navigation System Successfully Steers Simulated Planetary Landings

NASA engineers have successfully flown an autonomous navigation brain on a commercial drone, demonstrating the advanced sensor-fusion technology that will steer future uncrewed landers to the lunar surface. The system, tested extensively in the California desert, maps terrain and avoids hazards in real time without any human intervention from the ground.[1][4]

The Safe and Precise Landing – Integrated Capabilities Evolution (SPLICE) payload completed its simulated lunar descent maneuvers aboard an Alta-X drone. The August 27, 2026, flights took place over the expansive Rogers Dry Lake near NASA's Armstrong Flight Research Center in Edwards, California, marking a major milestone for the project.[1][2]

SPLICE is not a lander itself, but rather the sophisticated guidance and control package that future spacecraft will carry to the surface. It integrates terrain-relative navigation, hazard detection, and precision guidance into a single hardware and software suite designed to operate entirely on its own during the critical final minutes of flight.[1][3]

Mapping the descent

During a descent, the payload uses light-detection and ranging (LIDAR) sensors alongside high-speed optical cameras to scan the ground below. It builds high-resolution digital elevation maps on the fly, comparing the real-time topography against pre-loaded orbital data to determine its exact position, altitude, and velocity with pinpoint accuracy.[2][4]

The software continuously evaluates the targeted touchdown zone for boulders, deep craters, or steep slopes that could tip a spacecraft upon contact. If the primary site is deemed unsafe, the system autonomously calculates a new trajectory and diverts the vehicle to a better location, all within fractions of a second.[2][3]

AI

This autonomy solves a fundamental physics problem for interplanetary exploration: the absolute speed of light. Communication delays make real-time remote piloting impossible for anything beyond low Earth orbit, forcing spacecraft to manage their own final descents through complex, unmapped environments without waiting for instructions from Earth.[1][3]

A radio signal takes about 1.3 seconds to reach the Moon, and anywhere from four to 24 minutes to reach Mars, depending on the current orbital alignment. By the time mission control sees a hazard on their screens and sends a correction, a remotely piloted lander would have already crashed into it.[1][4]

The Artemis requirement

Previous Mars missions, like the Perseverance rover, utilized early versions of terrain-relative navigation to avoid hazards during their automated sky-crane descents. SPLICE represents the next generation of this concept, offering significantly higher resolution mapping and the ability to divert over much larger distances to find a safe haven.[3][4]

The technology is a direct and urgent requirement for NASA's Artemis program, which aims to establish a sustained human presence on the Moon. Future missions will need to deliver uncrewed cargo, habitats, and power stations to precise locations before astronauts ever arrive in orbit to use them.[1][3]

These automated payloads must land within meters of pre-positioned assets to be useful to the crew on the ground. The Apollo-era approach of astronaut eyeball navigation and manual stick flying cannot support uncrewed cargo drops, requiring a fully automated solution that guarantees pinpoint accuracy every single time.[1][4]

The challenge is heavily compounded at the lunar south pole, the primary target for the upcoming Artemis missions. The region features extreme lighting conditions, with long, deep shadows that can easily obscure massive rocks or steep craters from the view of orbiting cameras, making real-time surface scanning essential.[1][3]

Accelerating the test cycle

Researchers at NASA's Johnson Space Center in Houston spent nearly a decade developing the SPLICE hardware and software architecture from the ground up. They partnered with aviation specialists at Armstrong to integrate the complex package onto the Alta-X drone for rigorous atmospheric testing in the California desert.[1][2]

Using a commercial drone like the Alta-X is a deliberate strategy to accelerate the testing cycle and reduce overhead. Rather than relying on expensive sounding rockets or helicopter drops for every iteration, engineers can run dozens of flight profiles in a single week at a fraction of the cost and time.[1][4]

The flat, featureless, and geologically varied terrain of the Mojave Desert provides a highly practical analog for barren lunar landscapes. Preliminary flights conducted in June and July 2026 successfully verified communications, data handling, and telemetry between the payload and the drone's flight controllers.[1][2]

"This flight test is the culmination of almost a decade of work for the SPLICE project," said NASA project manager Davis Adams following the successful demonstration. He confirmed that the drone maneuvers validate the core algorithms required for planetary touchdowns, proving the math works in the real world.[1][2]

Adams noted that the results apply directly to terrestrial, lunar, and Martian descent scenarios without requiring fundamental changes to the software. The algorithms that process sensor data and command the drone's rotors are the exact same ones that will eventually throttle the descent engines of a lunar lander.[1][2]

Commercial technology transfer

NASA does not intend to keep the autonomous navigation technology proprietary or locked inside the agency. The agency plans to add the SPLICE guidance system to its software catalog this fall, making the code available to the broader American aerospace industry for commercial use and adaptation.[2][4]

AI

Commercial space companies are already evaluating the package for their own precision landing and autonomous surface mobility projects. This transfer of technology could significantly accelerate the development of private lunar landers competing for NASA's Commercial Lunar Payload Services contracts, lowering the barrier to entry for new firms.[2][3]

With the Alta-X drone flights now complete, the engineering team is preparing for a full-scale test at Voyager Technologies in Mojave, California, later this month. That test will push the integrated system closer to spaceflight readiness by simulating much more aggressive descent profiles and higher speeds.[1][2]

Translating to spaceflight

While the drone tests prove the software can process sensor data and command maneuvers in Earth's gravity, translating that performance to space remains a significant hurdle. The system must eventually operate in a hard vacuum, under reduced gravity, and aboard a violently vibrating rocket-powered vehicle.[1][4]

Spacecraft computers are notoriously slow compared to terrestrial hardware, as they must be heavily shielded against cosmic radiation to prevent fatal errors. The SPLICE algorithms must be efficient enough to process high-resolution LIDAR data in real time on these constrained, legacy-style processors without lagging.[3][4]

The ultimate goal is to integrate the SPLICE suite into the flight computers of actual lunar and Martian landers currently in development. Until that integration occurs, the system remains a highly capable prototype operating within a terrestrial testbed, proving the math before risking the billion-dollar hardware.[1][3]

The successful demonstration marks a critical transition from theoretical guidance models to field-validated navigation. It provides the foundational capability required to safely access the most scientifically valuable, and hazardous, regions of the solar system without human intervention, ensuring the next generation of explorers has the supplies they need.[1][4]