05 · Roadmap
Gaps, framed honestly
Strong sim + modular software; a real-robot product still needs hardware, vision, and trained skills. This is the v0.3.0 launch snapshot.
Done in repo v0.3.0
| Area | Status |
|---|---|
| Modular stack + World Model | Done |
| Mock + MuJoCo sim | Done |
| CLI (interactive + demos) + demo GIF | Done |
| Language (rule / BitNet / HTTP) | Done |
| Planner, policy, safety | Done |
| ROS2 adapter (dry-run) | Done |
| Vision / tracker interfaces | Done |
| Benchmarks + CI + Docker | Done |
| CONTRIBUTING, issues, CHANGELOG, release notes | Done |
Still open
| Priority | Gap |
|---|---|
| P0 | Live robot drivers (ROS2/SDK), calibration |
| P0 | Production detectors, pretrained YOLO adapter done (scene.backend: yolo, optional ultralytics); depth-based 3D pose still open |
| P1 | Diffusion policies + LIBERO-style eval (BC-trained MLP on randomized mock pick: done, kinetic train-bc, trained 100% vs random-init 0%) |
| P1 | Real IK: numeric DLS backend done (control.ik_backend: dls); MoveIt/pinocchio integration for real arms still open |
| P2 | Strong instruct LLM default; mobile base nav |
| P2 | Edge deploy (Jetson / TensorRT) |
| P3 | Safety certification, multi-arm / humanoid |
Strong sim + modular software; real-robot product needs hardware, vision, and trained skills.
This roadmap is the honest gaps list for Kinetic. Questions or pilots: contact DifrasAI.