NASA researchers have tested a compact 3D body scanner on printed manikins representing body shapes before spaceflight and on flight day 15. The prototype reconstructed a 47 mm difference in stature between models designed to differ by 46 mm. The ground demonstration examines a measurement approach intended for spacecraft, where large scanning booths and controlled movement around a subject are difficult to accommodate.
The work, described by K. Han Kim and colleagues from Leidos, KBR, and NASA’s Johnson Space Center, combines a handheld depth camera, coded visual markers, and software that joins separate views into a body model. It is a proof of concept for spaceflight anthropometry, the measurement of body size and shape, rather than a scanner already demonstrated on astronauts in orbit.
The system addresses changes caused by spinal elongation, fluid redistribution, and muscle loss. It also aims to capture neutral body posture, the posture a person adopts without gravity imposing the familiar requirements of standing or sitting. Both body dimensions and posture affect the relationship between a crew member and surrounding equipment. New Space Economy’s discussion of spacesuit sizing provides related background on accommodating different body shapes.
Why Conventional Scanning Is Difficult in a Spacecraft
The team considered several scanning approaches before developing its prototype. Booths, turntables, and arrangements of multiple cameras occupy space that vehicles may not provide. Some handheld and mobile-device methods require an operator to move slowly around the subject for several minutes, maintaining controlled motion as the images are collected. A floating subject and operator make that procedure difficult.
Camera alignment creates another problem. A scanning system needs to know how views relate to one another in three-dimensional space. Sensors that are bumped can lose their previously established alignment. Repeating a conventional calibration procedure inside a confined vehicle adds operational work.
An earlier station study conducted from 2013 through 2017 used two digital cameras with known optical characteristics. Their positions were calibrated against points on interior panels, allowing researchers to calculate body-marker positions. Ground-based scans of each subject were then repositioned to match the photographed posture. The new work seeks a more direct method for assembling a three-dimensional body measurement.
A Depth Camera and Coded Markers
The prototype uses an Intel RealSense D435 camera, which records both color images and depth. Depth measurements provide the distance information needed to place surface points in three dimensions.
Coded square markers, known as ArUco markers, are placed on the body and floor. Software recognizes each marker’s identity in the color image and associates it with coordinates from the depth measurements. Finding the same markers in separate views gives the system reference points for aligning those views.
The operator takes static captures from multiple positions instead of making one continuous, controlled circuit. An optimization algorithm adjusts each captured segment’s position and orientation to bring matching markers into agreement. Applying those adjustments joins the separate collections of surface points into a single model. This is marker-based self-calibration, rather than an absence of calibration information.
The assembled surface still contains noise and imperfections. The reconstruction software first fits a coarse body mesh, a connected surface defined by points, using a model trained on a database of human scans. That model constrains the result to plausible body shapes. A subsequent reconstruction increases the mesh resolution from approximately 600 vertices to about 40,000.
What the Manikin Tests Measured
The team printed two manikins based on earlier station-study results. Both represented first-percentile female stature, with one corresponding to a preflight body shape and the other to flight day 15. The nominal stature increase between them was 46 mm, approximately 3%.
The reconstructed models showed a 47 mm stature difference. Separately, the researchers compared each reconstructed surface with the computer-aided design model used to print its manikin. Root-mean-square surface error was 13.9 mm for the preflight model and 17.8 mm for the flight-day model. This measure summarizes discrepancies across the surface and gives greater weight to larger errors. Maximum errors were 27.8 mm and 35.6 mm, respectively.
The close agreement in stature difference and the larger surface errors describe different aspects of performance. The experiment detected the intended change between two known shapes; it did not establish millimeter-level accuracy across the entire body surface. Its test objects were stationary printed manikins.
The prototype still requires multiple captures with the subject remaining still. The team identifies form-fitting clothing printed with markers as its next development step, intended to reduce preparation and operating effort. It also proposes eventually incorporating validated sensor locations into vehicle or habitat interiors.
Useful Books Available on Amazon
