Skip to content

Repository files navigation

AISLE — Agentic In-Store Learning Environment

New contributor? Start with the AISLE contributor wiki for a source-linked project overview, architecture, use cases, extension guide, code map, research status, and known limitations.

Agentic auto-research for robot manipulation on open infrastructure: coding agents (Claude Code / Codex) compose and evolve typed dora-rs dataflows against a Genesis physics scene, with frozen verification/reset and hard safety structure. The claim under test: a typed dataflow substrate makes agentic robotics faster, safer, more auditable, and more reusable than script-level iteration — reproducible on a MacBook.

Full experiment design: docs/Project_AISLE_Experiment_Design.md. New to the repo? Start with docs/getting-started.md — and for the concepts behind it all (Physical AI, VLM/VLA/world models/WAMs, sim-to-real, agentic auto-research), docs/physical-ai-primer.md.

Status

This table is the single current status page (issue #142). Other overview pages link here; protocol and evidence pages may retain dated summaries for context, but must identify their snapshot and defer to this table on conflict. Status as of 2026-08-10, commit 0a19154. Each row states the verdict its committed evidence supports, with that evidence's own qualifications — a hypothesis with no admissible data says so rather than reading as progress.

Exact graph/manifest/CLI/ADR catalogs are generated, never hand-counted: docs/generated/project-inventory.md. Orientation for contributors: docs/contributor-wiki.md.

Milestone State
M0 — verified pharmacy-pick loop (SPEC 090) signed off; expert graph 0.98 pass@1 over 50 seeds, with the milestone replicate independently re-satisfying the gate
H1 — zero-shot composition measured, target not met: 40/40 schema-valid graphs, but 15% (claude) / 65% (codex) launch zero-shot; single dominant failure is uninstalled hub packages (analysis/h1/)
H2 — iteration to ≥90% claude arm met held-out (1.0 pass@1); codex arm 0.875 held-out at N=8 (one dropped), with dev-side evidence of a ≥0.9 system — see analysis/h2/ for the full verdict
H3 — skill accumulation (S1→S3 transfer) verdict PENDING (met: null, complete: false) — S2 and S3 both UNDECIDED. No admissible library-arm cell survived the integrity audit (repo, treatment or runtime drift); the wiped arm's clean cells never succeeded at S2/S3. A formal verdict needs an owner-accepted incomplete closure or a budget-corrected new campaign (analysis/h3/)
H4 — hot-swap vs relaunch iteration measured at T0, phase-randomized (ADR-h4 rev 2): hot-swap median iteration latency 32.4 s vs relaunch 41.8 s (ratio 1.29), n=6 per path, zero infra failures. Extremes overlap; no significance or equivalence claim at n=6. UNATTESTED dev measurement — makes no reproducibility claim (analysis/h4/)
H5 — zero wrong-object under free iteration holding on committed evidence: 0 wrong-object in 224/224 episodes across the three H2 campaign runs (analysis/h2/)
Retail suite S1–S3 (mobile, long-horizon) implemented: store scene, planogram verifier, mobility contract, S1 expert graph
Perception ladder L0/L1/L2 (TC-9) implemented: L0 oracle poses, L1 segmentation + depth (segmented-pose), L2 RGB identity + same-stamp sensor-depth geometry (l2-pose); the rung rides the graph and is asserted per run (--perception)
Realistic verifier (VER-5) implemented (src/aisle/verifier/realistic.py, OWLv2 + rules, CPU-pinned); ADR at docs/decisions/ADR-realistic-verifier.md
VER-6 verifier fidelity current VER-13 fusion recomputed over the same 31 recorded episodes: agreement 0.45, false SUCCESS 0.00 (0/6), false FAIL 0.68 (17/25). The preserved first, pre-amendment measurement was 0.29 / 0.00 / 0.88 (analysis/ver6-fidelity/; current recomputation in SPEC 040 VER-13). Conservative, not yet interchangeable with the oracle
CON-5 reproducibility on S1 original violation dispositioned: ADR-25 fixed and verified reset-anchored startup; ADR-26 defines full-episode outcomes as statistical under Metal noise. Issue #71 remains open for wall-coupled command/control timing and possible frozen-set retiming — per-seed outcome flips are not themselves a CON-5 violation

Quickstart

macOS arm64, Python >= 3.11, uv, and a Rust toolchain (for the dora CLI). Details and troubleshooting: docs/getting-started.md.

uv sync --extra sim        # plain `uv sync` REMOVES the sim extras
                           # NVIDIA host? use --extra cuda (GPU torch)
cargo install --git https://github.com/dora-rs/dora --rev cd597e705 dora-cli --locked
dora --version             # warns if CLI and python API revs drift

uv run pytest -m unit                            # fast, no simulator (~90 s)
uv run harness validate graphs/expert_t0.yaml    # typed-graph validation
uv run harness rollout --graph graphs/expert_t0.yaml --tier T0 \
    --episodes 2 --seeds 0..1 --no-idea-gate --env-baseline local
# With `uv sync --extra cuda`, add `--sim-extra cuda` (Linux only).

dora runs from source at the rev pinned in pyproject.toml [tool.uv.sources]; the CLI must be cargo-installed from the same rev. Never install with bare pip/conda.

Repository map

CLAUDE.md          development-agent contract (read first if you are an agent)
specs/             numbered specs with MUST IDs (000 = constitution)
TASKS.md           implementation order + kickoff prompts
registry/          capability schema + typed node manifests
graphs/            expert baseline dataflows (T0/T1 desk pick, S1 retail)
src/aisle/         scenes, bridge, verifier, reset, harness, mobility, nodes
harness CLIs       `uv run harness {validate,rollout,traces,report,skill,swap,probe}`
tools/             CI, trace_check, env_hash, campaign runners (H1/H2/H3/H4)
tests/             unit / sim / graph markers; every MUST cited by a test
analysis/          committed experiment findings (h1, h2, h3, h4, a1,
                   ver6-fidelity, s1-determinism, postmortems, transcripts)
docs/              guides, design doc, contributor wiki, decisions/ (ADRs),
                   generated/project-inventory.md (source-derived catalogs)
runs/              gitignored: traces, videos, run manifests

Structural counts (graphs, manifests, CLI commands, ADRs) are deliberately absent here — they went stale faster than anyone noticed. The generated appendix carries them and CI fails when it drifts.

How development works

Spec-driven, tests-first: specs define WHAT with numbered MUSTs; every MUST you implement needs a test citing its ID (tools/trace_check.py enforces this in CI); agents implement tasks from TASKS.md under the CLAUDE.md contract; humans review Class C paths (CODEOWNERS) and sign milestones. Gates before every commit: ruff format --check, ruff check, pytest -m unit, trace_check. Conventional commits; one concern per PR.

The local CI script also checks requirement traceability, the generated contributor inventory, and the committed frozen-environment hash.

The experiment's integrity rules are structural, not behavioral: the environment/verifier/reset set is hash-frozen (rollouts refuse to start on drift), oracle_state cannot be routed to policy nodes, all motion passes through the budget guard, and research agents operate under a separate contract (harness/CLAUDE.research.md) with idea-tree logging.

See docs/development-workflow.md for the full loop and docs/architecture.md for a tour of the system.

About

No description, website, or topics provided.

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages