SELIS V2 — ESA PECS proposal workspace
SELIS = Site Environment & Landing Infrastructure Simulator. Status: idea stage; guide and call package distilled. Last updated 2026-11-13. Anything marked [TBD] is undecided. Never fill a [TBD] with a guess; ask, or leave the marker in.
1. Purpose of this project
Turn a rough product idea (SELIS) into a complete, compliant, competitive Type A proposal for the 12th Bulgarian call under ESA's Plan for European Cooperating States (PECS), tender action 5-50197.
- Applicant (prime): Luminous Ltd, a Bulgarian legal entity, registered and approved on esa-star.
- Official closing: 26 November 2026, 13:00 Amsterdam time (14:00 in Sofia), per the ESA cover letter point 13. Ivaylo's working deadline is 25 November.
- Internal target: a submittable proposal by 23 November, leaving two days of margin.
Call constraints: Type A — Research and Preparatory Activities
Verified against the ESA cover letter
(call-docs/001.eli_5-50197.pdf).
- Aim: prepare participation in ESA optional programmes, or take the initial steps of a recurring product or service development for space with potential use on future ESA or commercial missions. Examples given: feasibility studies and demonstrators, conceptual design work, competitive landscape survey, user requirements, breadboard-based demonstrators.
- Start TRL at least 1; end TRL at most 3. TRL 3 is an analytical and experimental proof of concept of the critical functions; an integrated, validated system is TRL 4 and out of bounds.
- Expected duration 9 to 15 months.
- Firm fixed price not higher than EUR 200,000, fully funded, profit at most 8%.
- Overall weighted mark of at least 55 to be recommended.
- Inclusion of an end-customer is highly encouraged.
- Proposal at most 25 pages, in the ESA template, unchanged in structure and format.
- Non-Bulgarian entities: only where justified, not for core (IP-generating) work, subcontracts at most 20% of the price.
- No procurement of infrastructure or generic hardware or software tools.
- Duplication of ongoing or intended ESA activities is down-marked and can cause rejection.
- AI use must be disclosed: in cover letter point 9 for the proposal, and in section 2.4.3 for deliverables.
- From ESA's clarifications: the start TRL is stated for the system being developed, not its components; the proposal must substantiate the link to future missions and the path beyond TRL 3; cloud processing for the activity is an eligible cost; letters of interest from users may be annexed and are considered; no ESA staff member may be cited as adviser or endorser.
The full list of binding requirements is in
compliance-matrix.md. How to write each section is in
pecs-playbook.md.
Evaluation
criteria and weights (from
call-docs/ESA - TA Information.txt)
| Weight | Criterion |
|---|---|
| 40% | Clarity of technical objectives and requirements; quality of engineering approach and discussion of problem areas; quality and suitability of the programme of work; background and experience of the entity, including facilities; adequacy of key personnel |
| 30% | Prospects for use in ESA programmes; prospects for commercialisation and long-term benefit for the bidder and Bulgaria relative to the investment; consistency with programmatic objectives in the Cover Letter; credibility of the business plan; adequacy of current and targeted maturity |
| 25% | Management approach; rationale of the industrial organisation; credibility of cost estimate and planning |
| 5% | Compliance with administrative tender conditions, acceptance of the draft contract, compliance with IPR requirements |
Write and review every section against these four criteria.
evaluator-rubric.md breaks each one into questions.
2. What SELIS is
SELIS is a measurement-grounded, high-fidelity digital twin of the lunar surface. It unifies terrain, illumination, communications visibility, environmental data and user-placed infrastructure in one traceable, reviewable 3D sandbox.
Problem it solves: lunar data is not scarce, it is fragmented. Observations sit in different archives, instruments, resolutions, coordinate systems and toolchains. SELIS registers them in one spatial and temporal reference so analyses that are usually done separately (terrain, power, comms, thermal, operations) can be done together and in context.
No new flight mission is involved. The product is generated on Earth from existing mission archives.
How it is intended to be produced (design, not yet built)
This is the baseline approach from the RFI draft: a starting hypothesis, not a fixed design. Research may propose alternative data sets, methods or tools, and the Type A scope is not bound by these steps or by the outputs and use cases listed below.
- Regional terrain context from Kaguya/SELENE TC products (e.g. SLDEM2013) and LRO wide-area products.
- High-resolution site terrain from LROC NAC stereo imagery (about 0.5 m/pixel), co-registered to and constrained by LOLA geodetic control.
- Optional enhancement: data-driven stereo/3D reconstruction, super-resolution and feature segmentation where source data constrain it.
- Geometry from NASA/NAIF SPICE kernels (Sun, Earth, Moon, spacecraft, user-defined relay orbits).
- DEM-based ray tracing for time-dependent solar illumination, shadow, horizons, Earth visibility and Earthshine.
- Environmental layers where coverage exists (Diviner temperatures, rock abundance, ice-depth model products).
- Everything instantiated in a real-time, physically based 3D environment where landers, rovers and infrastructure can be placed.
Toolchain named so far: USGS ISIS, NASA Ames Stereo Pipeline, SPICE.
The four data classes (core design rule)
Every cell of every layer carries provenance and confidence, and belongs to exactly one class:
- Measured: mission observations. The primary reference.
- Derived: computed from measurements (slope, roughness, illumination, visibility).
- Inferred: AI or reconstruction output filling gaps or adding detail. Always tagged, never silently merged with measured terrain.
- Synthetic: procedural sub-resolution detail (centimetre-scale rocks, microtopography) for robotics simulation and synthetic sensor data. Never presented as ground truth.
This separation is SELIS's main credibility claim. Protect it in every text.
Outputs
- Layers: orthoimagery, DEM/mesh, slope/roughness/horizon, provenance/confidence, time-indexed illumination and shadow, Earth and relay visibility, Earthshine, optional thermal/geological/geotechnical, user asset and orbit layers.
- Formats: GeoTIFF/COG, LAZ/PLY, glTF/OBJ, CSV/Parquet, SPICE kernels, processing metadata; PDS4 labels on request.
- Reproducibility: versioned sources, fixed kernel sets, recorded processing parameters per release.
Use cases
Landing-site and exploration-zone assessment; hazard and terrain analysis; illumination and solar-energy availability; infrastructure placement; Earth and relay communications visibility; thermal analysis; rover traverse planning; surface-operations planning; synthetic data for perception and robotics development.
Current state (as of 2026-11-13)
Early preliminary feasibility work only. Do not describe it as more than this.
- NASA Ames Stereo Pipeline set up locally; DEMs produced from stereo image pairs to confirm the tooling works.
- Existing NASA DEMs and the LROC PDS archive explored.
- Rhino Grasshopper demos of photovoltaic potential computed on the generated DEMs.
- No integrated pipeline, no illumination engine, no 3D environment yet.
Long-term vision (beyond this proposal)
A site-agnostic system that ingests new mission data as it arrives and produces site packages on demand, sold to private companies, academia and potentially other agencies. The RFI draft estimated about 12 months to an MVP and 4 to 8 weeks per further site; treat those figures as unvalidated.
3. Important caveat about the source draft
The idea description comes from RFI/ESA_RFI_DRAFT.docx,
a fast, rough answer to a different ESA procedure (the RFI "Commercial
Lunar Data Products for Exploration Science"). Treat it as an idea
description, not a specification.
- The RFI response was submitted. Bulgarian entities turned out not to be eligible for that RFI; ESA replied that it is reviewing the project. Do not present this as ESA interest or endorsement.
- The RFI framed SELIS as a data product ESA would buy. PECS Type A funds research and preparatory work up to TRL 3. The proposal must be reframed around objectives, work packages, deliverables, schedule, cost and a start and end TRL.
- The full vision is far larger than a Type A activity (at most EUR 200,000, 9 to 15 months, TRL 3). A fundable slice must be chosen.
- Scope and implementation details in the draft may change. Flag conflicts rather than defending the draft.
- The draft's literature references and DOIs are unverified. Check each one against the publisher before it goes into any proposal text.
- The documents in
call-docs/are blank templates on purpose. Nothing from the RFI draft has been carried into them.
4. Team, partners and positioning
Key personnel
- Ivaylo Tsankov — Luminous Ltd. CV in
cvs/Ivaylo Tsankov CV.docx. [TBD: project role and time share] - Martin Ivanov — Master in architecture (New
Bulgarian University); PhD in civil engineering (University of Tokyo);
works in a laboratory at that university; involved in JAXA-related lunar
research on lava-tube habitats. CV to be added to
cvs/. [TBD: project role, time share, exact affiliation and project names from his CV] - Possible third person with AI experience — only if the chosen scope needs it. [TBD]
Use only what the CVs and Ivaylo state. Do not embellish credentials.
Partners and customer
- No subcontractors, academic partners or advisers planned. Add one only if critical to success.
- Candidate end-customer: a Lithuanian micro company developing solar
panels for space and the Moon, which will also apply to ESA. Still in
negotiation; name not yet shared. Lithuania is an ESA Associate Member,
so the call allows its participation where justified, for non-core work,
within 20% of the price if subcontracted. [TBD: name, and form of
involvement: letter of interest in an annex, or subcontract in a
customer role. ESA accepts both for Type A. See
pecs-playbook.mdsections 7 and 11.] - No other potential customers contacted yet.
IP and commercial position
- IP and licensing towards ESA are governed by the ESA contract
conditions in
call-docs/. - Everything else stays with Luminous Ltd.
- If the project succeeds, the product is sold to private companies, academia and potentially other agencies.
5. Open decisions
- Decided (2026-11-13): the Type A activity focuses on the lunar South Pole. [TBD: the specific site or sites within it]
- [TBD] Type A scope: which slice of SELIS, and what is explicitly out. To be derived from the Type A definition, the guide, the budget and the timeline, not from the RFI draft.
- [TBD] End goal and deliverables. Ivaylo's preference: a working demonstrator of one site with simulation and detailed analysis relevant to deploying a permanent base or a lander. Boundary from the call: it must be a proof of concept of the critical functions (TRL 3), each demonstrated and checked against real data, not an integrated validated system.
- [TBD] Start TRL at system level (1 or 2, given the current state), the evidence for it, and the evidence for the end TRL
- [TBD] Duration (9 to 15 months), price (up to EUR 200,000, fully funded), work-package structure (target four to five work packages)
- [TBD] Team composition: roles, time shares, whether an AI specialist or other expertise is needed; how each person is engaged and paid by Luminous Ltd, and where each will do the work
- [TBD] End-customer involvement and any further evidence of user need
- [TBD] Fit with ESA programmes and with Bulgaria's priorities under PECS
- [TBD] Competitors and prior ESA-funded work in this area
- [TBD] AI disclosure wording for cover letter point 9 and section 2.4.3, written at the end from what was actually done
- [TBD] Background IP: whether any existing Luminous work is declared
- [TBD] Whether to request the 35% advance payment that Luminous is entitled to as an SME
Record each decision in decisions-log.md with date and
reason, then update this file.
6. Rules for Claude in this project
Order of authority: the official call documents in
call-docs/, thenpecs-playbook.md(distilled from the guide inguide/), thendecisions-log.md, then this file, then the RFI draft. If two disagree, say so and follow the higher one.Do not invent facts: no made-up team credentials, customers, letters, performance figures, TRLs, costs or citations. Use [TBD] or ask.
Every factual claim about data sets, missions, competitors or markets needs a source that was actually checked. Give the link.
Keep the four data classes distinct in all wording. Never describe inferred or synthetic output as measured.
Write in plain technical English for a sceptical ESA evaluator who may read English as a second language. British English. Define each acronym on first use in every section. These rules are derived from ASD-STE100; do not claim STE compliance. a. One idea per sentence. Aim for 25 words or fewer; split anything longer unless splitting makes it less clear. b. One topic per paragraph, six sentences at most. c. Active voice with the actor named: "Luminous will validate the DEM against LOLA", not "the DEM will be validated". d. One term for one thing. Use the terms in rule 10 and in
glossary.md; do not vary them for style. e. No vague qualifiers ("significant", "robust", "state-of-the-art", "seamless"). Give the number, the comparison or the evidence, or cut the claim.Rules a to e apply in full to proposal sections and ready-to-paste sentences. Research briefs and internal notes need only d and e.
Respect the call's template: same headings, order and numbering, 25 pages at most. Use the page budget in
pecs-playbook.mdsection 4. Keep a running note of which proposal elements were generated or substantially modified with AI; the call requires this to be disclosed.One job per chat (one research topic, or one proposal section). End each chat by producing a short file to save into the project; chats do not share context otherwise.
Research chats follow
research-brief.mdexactly: one topic per chat, at most two pages, output saved asresearch/findings/<topic>.md.When reviewing, score against
evaluator-rubric.mdand be blunt. Lead with the weaknesses.Keep terminology fixed: "SELIS", "digital twin", "site package", "measured / derived / inferred / synthetic".
7. Project files
| Path | Content | Status |
|---|---|---|
CLAUDE.md |
This file. The master copy. README.md and
AGENTS.md are symlinks to it: always edit
CLAUDE.md, never the other two, and never replace them with
regular files. |
Done |
decisions-log.md |
Dated decisions and facts | Started |
research-brief.md |
Instructions and output format for research chats | Done |
glossary.md |
Terms and acronyms, one meaning each | Done |
sprint-plan.md |
Dated plan to the deadline, with status | Done |
call-docs/001.eli_5-50197.pdf |
ESA cover letter: activity types, limits, criteria, deadline | Added |
call-docs/002.ecc_5-50197_Fully Funded.pdf |
Draft contract for fully funded activities (applies to Type A) | Added |
call-docs/003.ecc_5-50197_Co-Funded.pdf |
Draft contract for co-funded activities (not used for Type A) | Added |
call-docs/004.etc_5-50197.pdf |
Tendering conditions | Added |
call-docs/005.ept_5-50197.docx |
Cover letter and full proposal template (blank) | Added |
call-docs/PSSA1_i5.xlsx, PSSA2_i5.xlsx,
PSSA8_i5.xlsx |
Cost forms: rates and overheads; price breakdown with Exhibits A and B; manpower and price per work package | Added |
call-docs/ESA - TA Information.txt |
Key acceptance factors and evaluation criteria as shown in esa-star | Added |
call-docs/clarifications/clar5-50197e1.pdf …
e9.pdf |
ESA clarifications 1 to 9. Official answers; they rank with the call
documents. Summarised in pecs-playbook.md section 11. |
Added |
guide/Essentials Guide to Bidding into ESA issue 1.0 - PECS Bulgaria.pdf |
Bid-writing guide, 182 pages | Added |
RFI/ESA_RFI_DRAFT.docx |
Original idea description | Added |
cvs/ |
Team CVs | Ivaylo added; Martin pending |
pecs-playbook.md |
How to write each section: distilled guide advice, checked against the call | Done |
compliance-matrix.md |
Every binding requirement of the call, with source and status | Done |
evaluator-rubric.md |
Criteria, questions, cross-checks, style check and output format for the evaluator chat | Done |
concept-note.md |
One-page Type A framing of SELIS | Next |
research/findings/*.md |
Competitors, data sets, users and needs, tech, feature opportunities | After concept note |
sections/*.md |
Proposal section drafts | Later |
8. Sprint plan
Official closing 26 November 2026, 13:00 Amsterdam time; working
deadline 25 November; internal target 23 November. The dated plan and
its status are in sprint-plan.md. Check it at the start of
a chat to see which step is current, and update the status column when a
step is finished.
9. Glossary
Terms and acronyms are defined in glossary.md. Use them
exactly as defined there, and add any new term to that file before using
it in proposal text.