HORUS v0.1.6-alpha

[whochangedmycode]

HORUS v0.1.6 Release Notes

Release Date: December 2025
28 commits since v0.1.5

---

Highlights

This release brings major improvements to simulation capabilities, ultra-low-latency networking, Python bindings, and cross-platform installation. Key highlights include:

• io_uring networking for 3-5µs latency on Linux
• sim3d production-ready with 17 sensor types and RL/ML integration
• Opt-in scheduler learning phase for better determinism control
• Universal installer with embedded/minimal profiles for Raspberry Pi, Jetson, and ARM SBCs

---

New Features

Ultra-Low-Latency Networking with io_uring

Complete io_uring implementation for Linux systems:

• 3-5µs network latency (matching Zenoh-pico performance)
• Zero-copy networking using registered buffers
• SQPOLL mode for kernel-side polling (can saturate 100Gb NIC on single core)
• Requires Linux 5.6+ (5.19+ for SQPOLL without root)
• Enable with io-uring-net feature flag

// Automatic detection - uses io_uring when available
let hub = Hub::new("sensor_data@192.168.1.100:9000")?;

sim3d - Production-Ready 3D Simulator

A complete 3D robotics simulator built with Bevy 0.15 and Rapier3D:

17 Sensor Types:

• RGB-D/Depth cameras with distortion modeling
• IMU (Inertial Measurement Unit)
• GPS/GNSS positioning
• LiDAR 3D scanning
• Event cameras (neuromorphic)
• Force/Torque sensors
• Tactile sensors
• Thermal cameras
• Sonar/Ultrasonic
• Radar
• Encoders
• And more...

RL/ML Integration:

• Curriculum learning system
• Adversarial training support
• Python bindings for RL agents
• Pre-built RL tasks: Balancing, Locomotion, Manipulation, Navigation, Pushing, Reaching
• 60 FPS visual / 100K+ steps/sec headless

Asset Support:

• URDF, MJCF, SDF import
• Gazebo model loading
• YCB object dataset integration
• FBX, OBJ, STL, COLLADA, USDZ formats

Physics:

• Soft body simulation (cloth, ropes, particles)
• Advanced constraint handling
• Rapier3D integration

sim2d Improvements

• Joint command support for manipulating kinematic chains
• Python API for 2D simulation scenarios
• Kinematics and inverse kinematics solvers
• Recording and playback functionality
• Metrics collection during simulation

Scheduler - Opt-in Learning Phase

The adaptive learning phase is now opt-in instead of always-on:

let config = SchedulerConfig {
    learning_phase: false,  // Disable for deterministic behavior
    ..Default::default()
};
let scheduler = Scheduler::with_config(config);

This provides better control for:

• Systems requiring deterministic timing
• Production deployments where profiling overhead isn't needed
• Hard real-time applications

Python Bindings - Network Support

Full network communication support in Python:

import horus

# Network-aware Hub (auto-detects local vs remote)
hub = horus.Hub("sensor_data@192.168.1.100:9000")

# Check if using network transport
if hub.is_network():
    print(f"Connected to remote endpoint")

# RouterClient for multi-robot communication
client = horus.RouterClient("robot_fleet@router.local:8080")

New Built-in Nodes

HORUS now includes 37+ production-ready nodes:

New in v0.1.6:

• KeyboardInputNode - Terminal keyboard capture (crossterm-based)
• JoystickInputNode - Gamepad/joystick support (gilrs-based)
  • Button mapping profiles (Xbox360, PlayStation4, Generic)
  • Axis calibration support

Categories:

Category	Count	Examples
Sensors	12	Camera, Depth Camera, LiDAR, IMU, GPS, Encoder
Actuators	8	DC Motor, BLDC, Stepper, Servo, Dynamixel
Safety	3	Battery Monitor, Safety Monitor, Emergency Stop
Communication	6	CAN Bus, Modbus, Serial, I2C, SPI
Navigation	4	Odometry, Localization, Path Planner, Collision
Control	2	PID Controller, Differential Drive
Input	2	Keyboard, Joystick

---

Improvements

Universal Installation Script

The installer now supports platform profiles for different deployment targets:

Profile	Packages	Best For
Full	All (sim2d, sim3d included)	Desktop, Workstation, Development
Embedded	Core + library, no heavy GUI	Raspberry Pi, Jetson, ARM SBCs
Minimal	Essential runtime only	Resource-constrained devices, CI/CD

Auto-detection for:

• Raspberry Pi (all models)
• NVIDIA Jetson
• BeagleBone
• Generic ARM SBCs

New Features:

• Smart progress bars with ETA calculation
• Automatic dependency detection and installation
• GitHub issue creation on build failure (with permission)
• Better error recovery for locked files
• Version-specific library caching

Dashboard/TUI Updates

• Replaced tui-nodes graph view with canvas-based visualization
• Improved responsiveness and visual feedback
• Network communication detection in monitor command
• Added horus profile command for system profiling

Documentation

• Enhanced search functionality in docs-site
• Fixed panel docking in sim3d editor
• Updated built-in nodes documentation

---

Breaking Changes

horus_mcp Extracted

The horus_mcp development tooling has been moved to a separate repository. If you were using the MCP server for development, install it separately:

# horus_mcp is now a separate tool
# See: https://github.com/softmata/horus-mcp

Topic Naming

Topic names now use . instead of / as separators internally. This is handled automatically but may affect debugging output:

# Old: /sensor/lidar/scan
# New: sensor.lidar.scan

---

Performance

Metric	v0.1.5	v0.1.6
Hub round-trip	~481ns	~481ns
Link round-trip	~248ns	~248ns
Network (io_uring)	N/A	3-5µs
sim3d headless	N/A	100K+ steps/sec

---

Bug Fixes

• Fixed docking panel behavior in sim3d editor
• Fixed verify.sh script issues
• Fixed degree viewpoint calculation when spawning MAP in sim3d
• Fixed discovery context handling
• Improved error handling in install script for locked files

---

Upgrading

# Clean install recommended
rm -rf ~/.horus/cache
cd /path/to/HORUS
./install.sh

Select your platform profile:

1. Full - Desktop/workstation development
2. Embedded - Raspberry Pi, Jetson, ARM SBCs
3. Minimal - CI/CD, resource-constrained devices

---

Dependencies Updated

• Bevy: 0.15 (graphics/simulation)
• Rapier3D: 0.22 (physics)
• PyO3: 0.27 (Python bindings)
• io-uring: 0.6 (Linux networking)

---

What's Next (v0.1.7 Preview)

• WebAssembly support for browser-based simulation
• ROS2 bridge improvements
• Additional RL environments
• GPU-accelerated perception nodes

---

Links

• Documentation
• Installation Guide
• Discord Community
• GitHub Issues

---

Full Changelog: v0.1.5...v0.1.6

---

This discussion was created from the release HORUS v0.1.6-alpha.