Advancing NASA's Next-Generation Optical Spacecraft Navigation System
Crowdsourcing robust algorithms to detect lunar crater rims for next-generation spacecraft navigation.
The Challenge
Crater rims are vital landmarks for planetary science and navigation. Yet detecting them in real imagery is tough, with shadows, lighting shifts, and broken edges obscuring their shape.
In pursuit of next-generation, terrain-based optical navigation, NASA is developing a system that uses a visible-light camera on a spacecraft to capture orbital images of lunar terrain and process the imagery to
Detect crater rims in the images
Match the craters to a catalog
Estimate the camera/vehicle position based on the identified craters
The focus of this project was the crater detection process. Natural imagery varies significantly in lighting and viewing conditions, which impacts the completeness of crater rims in the images. NASA sought a software function that could analyze such imagery and reliably fit ellipses along the edges of visible crater rims, across varying altitudes, viewing angles, crater densities, sizes, rim sharpness, lighting, shadowing, and contrast conditions, and in the presence of myriad non-crater features. Solutions also had to operate within strict CPU-only hardware constraints to be viable for spacecraft deployment.
The Solution
The Image-Based Crater Detection Challenge took a crowd-based innovation approach to develop robust, CPU-only algorithms for automated lunar crater detection. The challenge followed a structured lifecycle, including challenge definition, scoring configuration, dataset validation, public launch, submission, and automated evaluation, ensuring fairness, transparency, and reproducibility across all solutions.
Participants delivered software optimized for detection accuracy, reduced false positives, and efficient runtime and memory usage under strict hardware constraints. Upon completion, NASA received submission artifacts for the top-performing solutions, including full source code, installation and execution documentation, dependency details, standardized output formats, and accompanying technical white papers.
Challenges we ran:
1400+
Submissions
537
Participants from 27 countries
7
Winners
$55K+
In Prizes
The Impact
The challenge generated strong global engagement, attracting over 1,400 submissions from solvers across 27 countries, demonstrating the effectiveness of crowd-based innovation in sourcing diverse, high-quality technical solutions. Winning models exceeded performance expectations while complying with operational constraints, providing NASA with several viable approaches for further evaluation.
Several submitted solutions demonstrated significant performance improvements over baseline methods and are being evaluated within internal navigation research workflows to assess downstream performance benefits. Beyond the immediate task order, the results establish a validated foundation for future space exploration efforts, supporting both human and robotic missions operating in cislunar space.
Winners
1st place: motokimura
2nd place: kiminya
3rd place: apoplavsky
4th place: kukusi
5th place: number13
6th place: daisuke0530
7th place: tcghanareddy
Achieve high-quality outcomes with
Topcoder.
Achieve high-quality outcomes with Topcoder.