AEye Apollo LiDAR Selected for Lunar Outpost Pegasus Rover Moon Mission

AEye's Apollo software-defined LiDAR must survive 1000 g shock testing to ride on Lunar Outpost's Pegasus rover, and it has passed, clearing the sensor for a 2028 NASA launch under the Artemis program. The selection puts a hardened automotive-grade 3D LiDAR, rather than the one-dimensional ranging units used previously for docking maneuvers, onto a vehicle that must navigate the lunar surface autonomously.

Matt Fisch, CEO of AEye, described the design problem in an interview on the Visions podcast with host Jim Tatum: the hardware available off the shelf for space was docking LiDAR that measures distance to a single point, while NASA and Lunar Outpost needed autonomous-driving-grade 3D perception to characterize objects and terrain. Building a space-grade LiDAR from the ground up would take many years, so AEye proposed hardening its automotive product instead.

Why a MEMS scanner clears the launch bar

The heart of Apollo's scanning system is a MEMS device whose only moving part measures roughly one millimeter across. That geometry matters. Mechanical spinning-mirror LiDARs of the kind visible on Waymo vehicles carry motors and mirrors with real inertial mass, and they do not survive the 1000 g shock specification for launch and lunar landing, according to Fisch. The MEMS mirror has almost no inertial mass, and AEye tested it in vacuum with good results, he said.

Radiation hardening required changes to the electronics. Earth's magnetosphere shields terrestrial electronics from space radiation; the moon has no equivalent protection, and radiation wreaks havoc on laser-based electronics. Fisch said AEye moved the design toward fault-tolerant, space-grade electronics and passed the required testing.

Thermal cycling is the fourth hazard. Fisch cited temperature swings of 500 degrees Fahrenheit between shadow and sun, from roughly minus 400 degrees in a crater shadow to plus 250 degrees in sunlight, with near-instant transitions when a rover crosses a rock line. Lunar Outpost contributes rugged enclosures for that part of the problem.

What changes for navigation on the moon

Fisch estimated that 80 to 90 percent of the sensing task, producing the 3D point cloud consumed by the autonomous navigation system, is identical to terrestrial autonomous driving. The differences sit in interpretation. There are no highway maps, terrain reasoning dominates, and a camera cannot measure how far away a rock is, how rough the surface is, or what slope it presents.

The laser waveform itself carries material information, and AEye applies real-time AI processing to extract it. Fisch said the system can detect ice ahead on the surface — a capability with no equivalent in current terrestrial stacks, where robaxis confine themselves largely to Sunbelt states and clear weather. He tied the capability to AEye's Project Winter collaboration with the University of Toronto Robotics School, which targets detection of black ice, rain, and snow particulates.

Multiple-echo techniques, which take several shots of a scene and use AI to see through moving particulates, address lunar dust. Fisch said the LiDAR already sees better than a human in dust and fog.

Reprogrammability as the deciding factor

Fisch was direct about what won the contract: a unit launched to the moon is never serviceable. Apollo's MEMS chip allows shot-by-shot laser control and fully reprogrammable scan patterns, including zoom modes, with no hardware change. Mission parameters such as frame rates and resolution are, in his words, a good guess at launch, and the ability to alter how the sensor works through an over-the-space command sealed the deal.

The same reasoning transfers to vehicle manufacturers on Earth, who prefer to update sensors in the garage overnight rather than recall them. Fisch framed the mission's review and audit gauntlet as a proof point for the sensor's ruggedness in mining, autonomous trucking, and passenger vehicles, and predicted the moon-derived techniques will return to terrestrial driving as edge computing, including AI platforms from companies like Nvidia, advances.

The development leaves one question open for program managers: whether reprogrammable, OTA-updatable perception sensors become a baseline procurement requirement for every uncrewed lunar and planetary surface mission from 2028 onward.