Yantrika

v0.13.0 · 11 test suites, 700+ tests, all green · MIT license · hosted at yantrika.ai

An end-to-end fleet-operations platform for multi-vendor AMR (autonomous mobile robot) fleets — the operations layer for Physical AI. Robots (or the bundled simulator) speak VDA 5050 v2.1 over MQTT; a connector translates that into a shared Supabase schema; on top sit a live web console with a human approval gate, automatic incidents, predictive maintenance, real alert delivery, and an AI copilot grounded in your fleet's actual data.

Two ways to use it, one codebase:

Yantrika Cloud (hosted, free while in beta)	See the demo — no login · Create a free workspace · Start a throwaway live sandbox. Signing up provisions a private workspace (your own site_id, admin role, ingest keys to connect robots over VDA 5050 or REST, a one-click simulated fleet, role-scoped seats for your team). Contract: docs/WORKSPACES.md.
Self-hosted (MIT, forever)	The quickstart below runs the whole platform on your laptop with no cloud account; deploy/ has scripted AWS and Azure kits; supabase/0019_workspaces.sql turns self-serve workspaces on for your own deployment.

Quickstart — 3 commands, zero config

git clone <this repo> && cd yantrafleet
./install.sh                                  # Windows: powershell -ExecutionPolicy Bypass -File install.ps1
.venv/bin/python -m yantraops up --loopback   # Windows: .venv\Scripts\python.exe -m yantraops up --loopback

Open the console URL from the READY banner (it auto-opens your browser). That is the FULL platform — simulated fleet, auto-incidents, predictive maintenance, notifier, Sarathi copilot — against an embedded in-memory backend. No cloud, no keys, no robots.

Docs: docs/ARCHITECTURE.md (what we built and how — the comprehensive reference) · docs/index.html (overview, architecture, FAQ) · docs/RUNBOOK-WINDOWS.md (PowerShell runbook) · deploy/aws/ (run it on EC2) · docs/SECURITY.md (security posture & hardening) · docs/COMPLIANCE.md (SOC2/ISO/EU AI Act/VDA5050 mapping)

All components share one Supabase (PostgREST) backend and one robots row shape:

robots(id, vendor, status, battery, pos jsonb [x,y], speed, task_kind,
       health, motor_temp, tasks_done, fault_msg, updated_at)
alerts(id, sev, msg, src, tlabel, ack, created_at)
fleet_meta(id=1, writer_id, sim_min, throughput, updated_at)
robot_telemetry(id, robot_id, ts, battery, speed, motor_temp, status, pos)
incidents(id, sev, title, src, tlabel, state, impact, rca, fix, dur,
          created_at)  — written by the detector, read/patched by console
                         and copilot (rca/fix nullable)
commands(id, robot_id, cmd, status, requested_by, decided_by, ..., note)
missions(id, name, robots jsonb, state, prog, eta, created_at)
          — upserted by the simulator's rolling mission pool, read by console
maintenance_findings(id, robot_id, component, finding, rul_days, confidence,
          action, state, created_at, cleared_at)
          — written by the detector's --maintenance mode, read by console

以上为 README 的开头部分。其余内容请阅读原文。 查看完整 README ↗