Skip to content

Autonomy, GN&C, and Onboard Computing

Cross-cutting topic page for SST projects in spacecraft autonomy, guidance/navigation/control, and onboard computing. These projects push CubeSats beyond passive payloads toward self-reliant systems that can navigate, decide, and act without ground-in-the-loop.

Updated: session 37, 2026-04-14

The Swarm Autonomy Thread (ARC + Partners)

The most coherent technology lineage in all of SST: a decade of progressively more autonomous multi-spacecraft operations, all anchored at Ames Research Center.

Mission Year Size What it proved Outcome
10941 EDSN 2015 8×1.5U (Lost in Super Strypi failure) Launch failure
91369 Nodes 2016 2×1.5U Inter-satellite relay, autonomous networking flew
106824 V-R3x 2021 3×0.5U Radio ranging, mesh networking flew
106822 Starling 2023 4×6U Distributed autonomous ops: MANET, StarFOX, ROMEO, DSA flew
155355 Starling 1.5 2024 same Extended mission with STM software flew
155368 OSE-SAT 2024 hosted Containerized autonomy on D-Orbit ION, first SpaceROS flight flew (suggestive)
155367 PY4 2024 4×1.5U Cost-reduced swarm: LoRa ranging, 1-GPS-anchor, magnetorquer-only flew

Starling DSA firsts (confirmed via NTRS 20240004120, NASA press release Feb 2025): First fully distributed autonomous multi-spacecraft operations. First space-to-space comms for autonomous status sharing. First distributed reactive ops. First general-purpose automated reasoning on a spacecraft. First fully distributed automated planning across multiple spacecraft. Scalability study demonstrated feasibility up to 60 spacecraft in simulated lunar orbit (~100 tests). DSA PM: Caleb Adams (ARC).

Starling 1.5 STM with SpaceX (confirmed via NASA ARC press release): NASA partnered with SpaceX to test autonomous space traffic coordination. Starling CubeSats used SpaceX's Starlink screening service to detect a conjunction, accepted responsibility for maneuvering, and autonomously planned and executed an avoidance maneuver resolving a close approach with a Starlink satellite — first automated conjunction avoidance between different operators. U.S. Office of Space Commerce participated in validating the screening service. This extends the swarm arc from technology demos into operational space traffic management.

OSE-SAT — Docker-on-Orbit

155368 | ARC | TRL 3→7 | PI: Caleb Adams

Not a standalone spacecraft — an architecture. OSE-SAT is a hybrid-trust containerization framework that allows low-TRL autonomy software to run on-orbit in isolation from verified flight software. Docker-style Linux kernel features provide CPU/memory caps so a misbehaving container can't destabilize the host. First flight of SpaceROS (ROS2 adapted for spaceflight, developed jointly by Blue Origin and NASA).

Flew as hosted payload on D-Orbit ION vehicle. Confidence: suggestive (SPIE 2025 paper abstract + TechPort description; independent flight confirmation not verified).

The contribution: Starling proves autonomy algorithms work. OSE-SAT proves the deployment architecture works. Together they form the complete DSA stack.

PY4 — The Cost-Reduction Path

155367 | Carnegie Mellon | TRL 4→8 | PI: Zachary Manchester

4 × 1.5U CubeSats launched March 4, 2024 on SpaceX Transporter-10. Built on PyCubed open-source avionics (Python-programmable, integrates power/compute/comms/ADCS). Confidence: confirmed.

Core innovation is hardware elimination: one GPS anchor + LoRa two-way time-of-flight ranging reconstitutes the full swarm's orbital state. Magnetorquer-only ADCS (no reaction wheels). This is the cost-scaling path for constellation missions.

Demonstrated: mesh networking, inter-satellite ranging, range-based relative orbit determination, magnetorquer sun pointing, TID radiation measurements across 4 nodes.

Zachary Manchester — Associate Professor, CMU Robotics Institute. Director, Robotic Exploration Lab (REx Lab). PhD MIT AeroAstro. Founded KickSat (2011, Kickstarter-funded femtosatellite). Research: trajectory optimization, variational integrators, cislunar transfers. PyCubed is deliberately open-source — no commercial spinout. Confidence: confirmed.

Cislunar Autonomy Cluster

Advanced Space CAPS — 106812 + 106820

See Advanced Space org page for full writeup. CAPS autonomous navigation validated on CAPSTONE in NRHO. SBIR→SST→$72M AFRL Oracle pipeline.

SmallSat Steward — 155364

U Michigan | TRL 3→6 | PI: Oliver Jia-Richards | 2023-09 → 2025-09

Autonomous online learning and planning for safe inspection of cislunar stations. Based on the Dyna framework: reactive, integrated architecture for real-time model updates and trajectory planning. Design reference mission: exterior inspection of Gateway.

People chain: Jia-Richards came from MIT Space Propulsion Laboratory (Paulo Lozano's group) — the same lab that produced Accion Systems and the GPDM electrospray heritage. MIT SPL→U Michigan is a people chain connecting propulsion (Lozano) to autonomy (Jia-Richards). See U Michigan. Confidence: pending (no flight or commercial downstream identified).

CAPS-TONE (CAPSTONE navigation experiment) — 106820

Cislunar Autonomous Positioning System Technology Operations and Navigation Experiment. TRL 5→8. See Advanced Space. CAPSTONE was the first spacecraft to demonstrate autonomous cislunar navigation — the technology is foundational for Gateway GN&C.

SCALES — 155362

Cal Poly Pomona (MSI: HSI + AANAPISI) | TRL 3→? | 2024-01 → 2025-12

Spacecraft Compartmentalized Autonomous Learning and Edge-computing System. Autonomous onboard decision-making for resource-constrained spacecraft. No flight, no external lineage identified. Confidence: pending.

Asteroid Navigation ML — 95600

University of Arizona (Dante Lauretta) | TRL 3→5 | 2018-03 → 2020-10

ML algorithms for on-board asteroid shape model determination and spacecraft navigation. TechPort: Completed.

External lineage (confirmed — direct OSIRIS-REx feed-in): Lauretta is the PI of OSIRIS-REx (first NASA asteroid sample return, Bennu sample delivered Sep 2023; now continuing as OSIRIS-APEX toward asteroid Apophis, rendezvous 2029). The SST project ran concurrently with OSIRIS-REx proximity operations at Bennu (rendezvous Nov 2018; TAGSAM sample collection Oct 20, 2020 — same month SST closed out). The NTRS fact sheet (20180006633) carries funding number NNM10AA11C — the OSIRIS-REx New Frontiers contract — directly linking the SST ML work to the mission. OSIRIS-REx's final approach navigation used Natural Feature Tracking (NFT) with CNN-based image matching — the first time autonomous optical navigation guided a spacecraft to a planetary surface. Confidence: confirmed (NTRS 20180006633 funding number; NTRS 20170002016 lessons learned; contract NNM10AA11C appears in OSIRIS-REx mission documentation).

Autonomous Optical Navigation — 155359

Georgia Tech | TRL 3→6 | PI: John Christian | 2023-10 → 2025-09

Horizon-based optical navigation (OPNAV) instrument for autonomous cislunar and deep space navigation. 1U form factor.

John Christian — 2 SST projects 11 years apart from 2 different institutions (JSC 2013, GA Tech 2023). One of ~3 PIs to appear twice in SST. Former JSC engineer (2010–2012). 6 TechPort projects across 3 programs (SST, NIAC, STRG). AAS Fellow (2021). Algorithms in Artemis I Orion OPNAV system. After Lunar Flashlight propulsion failed, NASA transferred LONEStar mission to GA Tech; Christian's team performed first on-orbit heliocentric nav using only planet observations (~400 images, JAS 2024). GA Tech as SST alumni hub: both Christian and Lightsey moved from JSC to GA Tech. See Georgia Tech. Outcome: transitioned. Confidence: confirmed (Artemis heritage, AAS Fellow, LONEStar).

Surface Feature Nav — 106826

UT Austin (Brandon Jones) | TRL 3→? | Active (through 2027-08)

ML-based crater navigation and timing (CNT) for lunar orbit. STRG→SST pipeline confirmed (TechPort Advanced_To from [91512]). SCOPE-1 LEO CubeSat demo pending ~end 2026. See UT Austin. Confidence: pending (flight demo not yet launched).

Stanford SLAB Cluster (Simone D'Amico)

94049 Precision GNSS Nav — TRL 3→6, 2016-2018. Centimeter-level relative navigation in nanosatellite formations using GPS/Galileo/GLONASS. Peer-to-peer, decentralized. Confidence: confirmed (TRL 6 = hardware-in-the-loop demo).

95519 Autonomous Nanosatellite Swarming — TRL 3→5, 2018-2020. GNC algorithms for cooperative swarms using RF and optical sensors. Target: asteroid characterization. Research continued post-SST toward autonomous asteroid operations. Confidence: suggestive.

Other GN&C Projects (catalog only)

Project Lead TRL What
95540 Purdue 5→6 Distributed attitude control for deep space SmallSats
91591 Purdue 4→6 MEMS reaction control for picosats beyond LEO
91499 Northrop Grumman 5→7 Reaction Sphere — validated on parabolic flight (TRL 5-6) but no orbital demo or commercial product. Concept diffused to Tensor Tech (independent).
106813 Langley 3→5 Integrated precision ADCS
91330 U Maryland 3→4 Propellantless attitude control for solar sails using reflective control devices
91360 JSC 3→4 CubeSat autonomous rendezvous & docking software
91474 JSC 3→5 MEMS IMU swarms for precision navigation
90699 MIT 3→6 Smoothing-based relative navigation and coded aperture imaging

Onboard Computing

Project Lead TRL What Outcome
91661 GSFC (Brock LaMeres, Montana State) 5→6 Radiation-tolerant FPGA-based SmallSat computer (RadSat) commercialized
155363 Utah State 3→5 Low-power multi-FPGA cluster for edge computing, reprogrammable on-orbit no-visible-outcome
155362 Cal Poly Pomona 3→? SCALES autonomous learning edge computing no-visible-outcome

RadSat → RadPC → Lunar Surface (The Strongest University Pipeline)

91661 — GSFC-led but PI is Brock LaMeres (Montana State). RadSat FPGA computer: radiation-tolerant via FPGA-based TMR. RadSat-g deployed ISS July 2018; RadSat-u deployed ISS Feb 2020.

Downstream: LaMeres founded Resilient Computing (2020), licensed RadPC (2021). $2.78M federal funding (5 NASA + 1 DoD/SOCOM). RadPC flew on Blue Ghost 1 (CLPS), landed Mare Crisium March 2025 — met all mission objectives, longest-duration instrument of all 10 payloads, operated through Van Allen belt and into lunar night. First SST-connected university technology confirmed operating on the lunar surface.

See Montana State. Outcome: commercialized. Confidence: confirmed (Resilient Computing website, Blue Ghost 1 mission reports, USASpending).

Cross-Cutting Findings

  1. ARC is the institutional anchor for swarm autonomy. 7 SST missions over 10 years, absorbing a launch failure (EDSN), building from 2-node relay to 4-node distributed planning. The "product" is institutional expertise, not a commercial product. See Archetypes — Institutional Capability Builder.

  2. Cislunar is the emerging frontier. 7+ projects in this topic target Moon/cislunar destinations: CAPS (Advanced Space), SmallSat Steward (U Michigan), Autonomous OPNAV (GA Tech), Surface Feature Nav (UT Austin), SCALES, and the LunaNet PNT stack projects (UCLA, Caltech, SDSU). See PNT/Timing for the emergent architecture finding. This reflects STMD's pivot toward Artemis-supporting smallsat capabilities.

  3. The hardware-elimination thesis (PY4) vs. the sophisticated-hardware thesis (Reaction Sphere, precision ADCS) — the SST portfolio funded both. PY4's approach (magnetorquer-only, LoRa ranging, one GPS anchor) reached TRL 8 and flew. Northrop's Reaction Sphere was terminated. The cheaper approach won, at least for swarm applications.

  4. Academic-to-flight pipeline: John Christian (GA Tech OPNAV → Orion Artemis 1 + LONEStar), Simone D'Amico (Stanford SLAB → DiGiTaL → Starling StarFOX flight demo 2023), Dante Lauretta (U Arizona SST ML nav → OSIRIS-REx NFT sample collection 2020), Zachary Manchester (MIT → CMU → PyCubed → PY4 flight 2024), Brock LaMeres (Montana State → Resilient Computing → RadPC on the Moon 2025). SST funds academic researchers who later design flight systems — or the SST project is the flight system development. Lag time varies from zero (Lauretta: concurrent) to 12 years (LaMeres: RadSat 2013 → lunar landing 2025).

  5. JPL as integration node: Andrey Matsko (JPL) co-investigates both the Vahala cislunar clock (Caltech, [155361]) and the Wong LunaNet PNT module (UCLA, [106828]) — bridging two layers of the emergent LunaNet PNT stack. Rais-Zadeh (U Michigan phonon trap, [91596]) moved to JPL to lead the MEMS group. JPL appears to be the recipient institution for SST university research — it pulls in academics, and their technology, into NASA's in-house capability base.

  6. Convergence events. Multiple SST people chains converge on downstream missions: VISORS (4+ SST chains), SWARM-EX (3 chains), GPDM (3 chains, first intra-NASA convergence). See VISORS, SWARM-EX, GPDM surprises.

Cross-references