AI-BASED ROBUSTNESS TESTING · AVAILABLE NOW
ProximaAssure is ready to test your GNC and closed-loop control, including computer vision, navigation filters and their interactions.
Find more corner cases, reduce manual testing effort and reproduce failures before flight. OPL also provides simulation, visual navigation and mission-specific GNC integration.
Orbital servicing concept illustration
UK Space Agency Accelerator Fusion: Connect with Capital
UNITED KINGDOM / 2026WHAT WE BUILD
Orion Proxima develops AI-based robustness testing for spacecraft autonomy. ProximaAssure works with your existing algorithms. ProximaSim, ProximaVision and our mission-specific GNC kit support teams that also need to develop or integrate the autonomy itself.
PRIMARY PRODUCT · AVAILABLE NOW
AI-based robustness testing for GNC and closed-loop control, including computer vision, navigation and control interactions.
SUPPORTING PRODUCTS · DEVELOP AND INTEGRATE
Design your GNC with mission dynamics and realistic imagery, then generate C code for NASA cFS/cFE integration.
Explore ProximaSimDetection, tracking, relative pose estimation and ML-augmented navigation for onboard integration.
Explore ProximaVisionA mission-specific software package combining perception, navigation, guidance and control for close-proximity operations.
Explore the GNC kitPRIMARY PRODUCT / PROXIMAASSURE
Individual subsystems can pass their tests while the closed loop still fails. ProximaAssure searches for combinations of perception errors, navigation uncertainty, delays and control responses that expose those failures.
AI-based search targets challenging combinations of uncertainty, sensor degradation and control errors. In OPL’s internal benchmarks, it identified more than five times as many edge cases as Monte Carlo testing.
Reduce time spent creating and searching test scenarios, so engineers can focus on investigating failures. Connect your existing algorithms through a black-box interface without sharing source code.
Replay the conditions behind a failure and check subsequent software releases against the same scenario. Extend selected tests to representative onboard hardware through a scoped HIL setup.
Integration uses an executable, agreed interface or customer-operated test endpoint, with the outputs needed to evaluate behaviour.
01 / DESIGN · PROXIMASIM
A modular guidance, navigation and control (GNC) design suite that connects spacecraft dynamics, realistic sensor imagery and your autonomy algorithms.
02 / NAVIGATE · PROXIMAVISION
Computer vision and navigation software for close-proximity missions. Combine target detection and tracking with relative pose estimation and ML-augmented navigation filtering.
Target adaptation, processor compatibility and performance are evaluated for your mission.
Request a ProximaVision demoSUPPORTING PRODUCT / GNC KIT
A mission-specific GNC software integration package for close-proximity operations, bringing together perception, relative navigation, guidance and control.
Configuration, delivery scope and verification requirements are agreed per programme. This is a software integration offering; flight qualification remains mission-specific.
Discuss a GNC kit for your mission04 / SOFTWARE TO HARDWARE
Search broadly in software. Take the most valuable scenarios onto representative onboard computing and test interfaces to investigate integration effects.
SIL / EXPLORE THE SCENARIO SPACE
Run your software against simulated mission dynamics and sensor inputs. Explore uncertainty and fault combinations before committing time to hardware tests.
ProximaAssure varies scenarios and evaluates outcomes.
PIL / CHECK THE TARGET PROCESSOR
Execute compiled algorithms on a representative processor. Examine execution time, resource use and numerical behaviour against the software baseline.
Compare behaviour with the software baseline.
HIL / EXERCISE THE INTEGRATION
Connect representative onboard hardware to a real-time mission simulation through agreed interfaces. Replay selected scenarios to examine timing, data exchange and recovery behaviour.
Reproduce selected failures and rerun after a fix.
HIL integration scoped to your programme. Hardware, interfaces, timing requirements and any physical sensor stimulation are agreed before testing.
Discuss HIL integration05 / THE LAST MILE
The last mile is where sensing, estimation and control must work together. Our core focus is close-proximity spacecraft operations, with related development applications in robotic interaction and precision touchdown.
Design approach trajectories, estimate relative pose and verify hold points, alignment and abort responses.
GNC · RELATIVE NAVIGATION · ABORT LOGICDevelop target tracking and relative navigation for inspection fly-arounds and servicing under challenging illumination.
PERCEPTION · TRACKING · ESTIMATIONDevelop perception and precision alignment for robotic interaction and refuelling interfaces, with mission-specific contact models and testbeds.
ALIGNMENT · CONTROL · RECOVERYExplore terrain-relative navigation and terminal descent in simulation, then investigate perception and control behaviour through SIL/HIL testing.
TERRAIN NAVIGATION · DESCENT · TOUCHDOWNApplication scope is agreed per programme. Landing and contact-rich robotic operations require dedicated models, hardware integration and validation.
06 / ASSURANCE EVIDENCE
A useful failure is one your engineers can understand and reproduce. Evaluate ProximaAssure against your baseline, your test budget and your mission requirements.
Compared with Monte Carlo in OPL’s internal testing, with a broader range of failure cases identified.
An internal result, not a guarantee for every system. Ask us about the test setup, baseline and definition of an edge case.
Request benchmark detailsDefine the uncertainty ranges, baseline, run budget and computational cost.
Assess distinct failure modes, severity, reproducibility and time to discovery.
Retain the scenarios and configurations needed to check the next version.
Targeted failure search complements conventional verification. Finding a failure does not, by itself, establish its probability in flight or certify a system.
07 / ORION PROXIMA
Orion Proxima Ltd develops simulation, visual-navigation and verification software for precision spacecraft operations. We combine machine learning, robotics, computer vision and GNC expertise to help mission teams design the autonomy, integrate it onboard and test how it behaves.
More than 4.5 years working and consulting on RPO missions, including experience at Astroscale.
PhD-level computer vision and machine learning, with over 15 years of ML and robotics experience.
Selected for the UK Space Agency Accelerator’s Fusion: Connect with Capital programme.
Experience refers to the team’s prior work; it does not imply flight qualification of OPL products.
BEFORE WE TALK
Yes. ProximaSim brings mission dynamics, sensor models and realistic imagery into a modular guidance, navigation and control design environment. Connect perception, navigation filters, guidance and control algorithms, then evaluate the complete closed loop under mission-specific conditions.
Yes. ProximaSim supports C-code generation for GNC applications, with code prepared for integration into NASA’s core Flight System (cFS), using the core Flight Executive (cFE). This connects simulation-based development with your onboard flight-software workflow.
Target hardware, operating system, interfaces and timing requirements are agreed during integration. Mission-specific verification and qualification are still required before flight.
No. The starting point is your existing system. We agree the interfaces, scenario controls and outputs needed for a testing campaign. ProximaSim and ProximaVision are optional modules.
Source-code disclosure is not required for black-box testing. We need a way to run the system, supply inputs and inspect relevant outputs. The executable or customer-operated endpoint, data access and confidentiality arrangements are agreed during scoping.
We scope one mission or autonomy function around your needs: a simulation and GNC-development workflow, a visual-navigation evaluation, or an AI-driven robustness campaign. We agree the inputs, integration requirements and measurable outputs before work begins.
We can discuss a processor- or hardware-in-the-loop setup around your target hardware. Compatibility depends on its interfaces, software environment and real-time requirements. We agree the configuration before the campaign.
TELL US WHAT YOU ARE BUILDING
Book a 20-minute ProximaAssure demonstration. Bring your GNC or closed-loop testing challenge and see how AI-based search can help. We can also discuss Sim, Vision and GNC integration for your mission.
hello@orionproxima.com