[gabrielfrasantos]

Adds spec files (implementation.md, tests.md, explanation.md) for the planning family under roadmap/planning/, next to the existing W8–W10.

• P1 PrimitiveDistance — closed-form signed distances (negative means penetration) between points, segments, spheres, capsules and oriented boxes, treated as core plus radius. Uses Ericson's segment–segment routine, including its parallel and zero-length branches.
• P2 PotentialField — attraction that is quadratic near the goal and capped beyond d*, plus Khatib repulsion from proximity readings. The output is a clamped task-space velocity. Local minima are detected and reported, not hidden.
• P3 GridSearch — A* and Dijkstra on 2-D and 3-D grids.
  • Per-cell costs come from a callable, so E2's map plugs in without copying.
  • An indexed binary heap with decrease-key; 16-bit indices on grids under 65 535 cells.
  • The octile heuristic is scaled by the minimum cell cost, which keeps it admissible.
• P4 SelfCollision — a capsule per link, joint to joint, from M30's FrameChain. An allowed-collision matrix skips adjacent links by default. The check returns the minimum distance and the pair that attains it, so it also feeds P2, P6 and H2.
• P5 ConvexCollision — GJK on shape cores with the radii subtracted (exact for rounded shapes), plus a bounded EPA for overlaps. Covers spheres, capsules, boxes and bounded hulls.
• P6 RrtConnect — fixed node pools, with the sampler and validity test injected by the caller, plus random shortcutting. Runs are reproducible for a given seed.

ROADMAP: the P entries describe the designs above, and the intro now lists planning as specced. roadmap/README.md: P1–P6 added to the planning index.

Checks. Every reference value comes from a Python reference implementation:

• P1: closed forms against brute-force sampling.
• P3: A* against Dijkstra on wall and weighted maps.
• P4: folded-arm configurations.
• P5:
  • GJK against P1 and an exact box–box distance.
  • EPA against separating-axis depths: 300 random pairs, 500 deep overlaps and 634 axis-aligned overlaps with tied coordinates.
• P6: three seeds plus shortcutting.

That work produced two rules, now written into the specs as normative:

• EPA: orient faces from an interior point and count coplanar faces as visible. Without the first, axis-aligned boxes gave a depth of 0.150756 instead of 0.5; without the second, identical boxes produced NaN.
• RRT-Connect: the sampler must not correlate consecutive draws. An in-order base-2 Halton sequence stalled both trees at 19 nodes after 2000 iterations, while a seeded xorshift32 found paths in 78–104 iterations.

No upstream (numerical-toolbox) prerequisites are needed.

---

Generated by Claude Code

[github-actions]

✅⚠️MegaLinter analysis: Success with warnings

Descriptor	Linter	Files	Fixed	Errors	Max errors	Warnings	Elapsed time
✅ ACTION	actionlint	7		0		0	0.55s
✅ CPP	clang-format	35	0	0		0	0.63s
✅ CPP	cppcheck	35		0		0	0.82s
✅ DOCKERFILE	hadolint	1		0		0	0.55s
✅ JSON	jsonlint	8		0		0	0.1s
✅ JSON	prettier	8	2	0		0	0.4s
✅ MARKDOWN	markdownlint	182	0	0		0	2.19s
✅ MARKDOWN	markdown-table-formatter	182	0	0		0	0.21s
⚠️ SPELL	lychee	216		1		0	2.02s
✅ YAML	prettier	12	0	0		0	0.46s
✅ YAML	yamllint	12		0		0	0.47s

Detailed Issues

⚠️ SPELL / lychee - 1 error

📝 Summary
---------------------
🔍 Total..........135
🔗 Unique..........71
✅ Successful.....124
⏳ Timeouts.........0
🔀 Redirected.......5
👻 Excluded.........9
❓ Unknown..........0
🚫 Errors...........1
⛔ Unsupported......1

Errors in .github/workflows/static-analysis.yml
[403] https://binaries.sonarsource.com/Distribution/sonar-scanner-cli/sonar-scanner-cli-$ (at 37:21) | Rejected status code: 403 Forbidden

Hint: Followed 5 redirects. You might want to consider replacing redirecting URLs with the resolved URLs. Use verbose mode (`-v`/`-vv`) to see redirection details.
Hint: You can configure accepted/rejected response codes with `-a` or `--accept`

Notices

⚠️ Your configuration references items that have been removed from MegaLinter and are ignored: REPOSITORY_GITLEAKS, REPOSITORY_KICS. See Removed linters to find their replacements.

See detailed reports in MegaLinter artifacts

Your project could benefit from a custom flavor, which would allow you to run only the linters you need, and thus improve runtime performances. (Skip this info by defining FLAVOR_SUGGESTIONS: false)

• Documentation: Custom Flavors
• Command: npx mega-linter-runner@10.1.0 --custom-flavor-setup --custom-flavor-linters ACTION_ACTIONLINT,CPP_CPPCHECK,CPP_CLANG_FORMAT,DOCKERFILE_HADOLINT,JSON_JSONLINT,JSON_PRETTIER,MARKDOWN_MARKDOWNLINT,MARKDOWN_MARKDOWN_TABLE_FORMATTER,SPELL_LYCHEE,YAML_PRETTIER,YAML_YAMLLINT

Show us your support by starring ⭐ the repository

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud