Name: HORUS
Availability: InStock

Recipes

Copy-paste-ready patterns for common robotics tasks. Each recipe is a complete, self-contained program you can adapt and ship. Unlike tutorials (which teach step-by-step), recipes are pure working programs with inline safety annotations.

Every recipe includes:

A complete horus.toml manifest
Full source code with all imports and entry point
Inline // SAFETY: / # SAFETY: and // IMPORTANT: / # IMPORTANT: comments for critical patterns
Expected terminal output

Start Here

Recipe	What It Builds	Key Patterns
Real Hardware	I2C IMU + serial motors with real crates.io/pip libraries	Complete hardware path, no stubs

Rust Recipes

Recipe	What It Builds	Key Patterns
Differential Drive	2-wheel robot: `CmdVel` to motor commands	Actuator safety, shutdown, execution order
IMU Reader	100Hz IMU sensor with orientation publishing	Sensor node, `#[repr(C)]`, zero-copy
PID Controller	Generic PID loop with configurable gains	Control theory, rate-based RT
LiDAR Obstacle Avoidance	Reactive velocity from `LaserScan`	Sensor fusion, safety zones
Servo Controller	Multi-servo bus with safe shutdown	Multi-actuator, ordered shutdown
Multi-Sensor Fusion	IMU + odometry state estimation	Multi-topic aggregation, caching
Emergency Stop	E-stop monitor with safety state	Safety-critical, `Miss::SafeMode`
Telemetry Logger	Log topics to file via async I/O	`async_io()`, non-blocking file writes
Transform Frames	Coordinate frame tree with sensor mounts	`TransformFrame`, static/dynamic frames, `FrameId` caching
Record & Replay	Record execution, replay, mixed replay for regression	`.with_recording()`, `replay_from()`, `add_replay()`, CLI

Python Recipes

Recipe	What It Builds	Key Patterns
Differential Drive (Python)	2-wheel robot with kinematics + odometry	`CmdVel`, RPM clamping, safe shutdown
IMU Reader (Python)	100Hz IMU sensor with orientation	`horus.Imu`, typed topics, NaN validation
PID Controller (Python)	Generic PID with anti-windup	`horus.dt()`, integral clamping, gain tuning
LiDAR Avoidance (Python)	Reactive velocity from `LaserScan`	Three-zone split, safety stop, `CmdVel`
Servo Controller (Python)	Multi-servo bus with safe shutdown	`ServoCommand`, `JointState`, joint limits
Multi-Sensor Fusion (Python)	IMU + odometry complementary filter	NumPy, multi-topic aggregation, angle wrapping
Emergency Stop (Python)	Safety monitor with fail-safe	`EmergencyStop`, debounce, stale detection
Telemetry Logger (Python)	Async CSV logging	`async def tick`, `aiofiles`, non-blocking I/O
Coordinate Transforms (Python)	Frame tree with sensor mounts	`TransformFrame`, static/dynamic, point conversion
Python CV Node	Python computer vision with `horus.Node`	Python API, NumPy integration
Record & Replay (Python)	Record execution, replay and diff via CLI	`recording=True`, CLI replay, export to CSV

How to Use a Recipe

horus new my-robot -r
cd my-robot

Copy the horus.toml and src/main.rs from any recipe, then:

horus run

Conventions

Execution order: Publishers run before subscribers (lower .order() first)
Safety: Every actuator node implements shutdown() that zeros outputs
recv() every tick: Always call .recv() on every subscriber, every tick — even if you discard the value
No sleep(): Never use std::thread::sleep() — use .rate() on the scheduler
No blocking I/O in tick(): Use .async_io() for file/network operations