Willow 5 Runtime SDK (C++)

The Willow 5 Runtime C++ SDK is a dependency-free, header-only inference library designed for sub-millisecond execution of Zero-Shot Action Recognition directly on IoT, VR/AR Edge, and Robotic platforms.

While the Willow Cloud Oracle is used to train and generate proprietary Action Models, this C++ SDK allows you to execute those models locally in embedded environments. It acts as the high-performance bridge between Willow's biomechanical intelligence and your hardware.

Core Capabilities

• Dual Spatial Engines: Natively decodes V40 (Scale-Invariant) and V41 (Absolute Physics/Metric) signatures.
• Object Tracking (Node 76): Natively process human-object interactions (tools, balls) in a unified 3D metric space.
• Zero-Allocation Memory: Uses internal ring-buffers to guarantee deterministic execution times (< 1ms) without Heap fragmentation.
• Sim-to-Real Retargeting: Convert Computer Vision coordinates into ROS (Z-Up) and Unity formats natively.
• Edge Physics: Calculate Jerk, Acceleration, and Power without cloud latency using the ported Da Vinci engine.
• DRM-Compliant: Models can load securely into ephemeral RAM via libcurl to protect Intellectual Property.

---

Spatial Engine Modes (New in Q2 2026)

The C++ SDK automatically parses the binary header of your .int8 model to determine its mathematical routing:

• Universal Models (Scale-Invariant): Evaluates structural form and sequence. Distances are mathematically normalized by torso length. Expects a 75-point generic coordinate Skeleton.
• Physics Models (Metric Space): Evaluates exact volumetric paths and physical velocities (m/s). Normalization is bypassed. Expects coordinates to be provided in absolute Meters, utilizing a 76-point Skeleton where Index 75 represents the tracked reference object.

---

🚀 Quickstart

The SDK is completely header-only. Simply include include/willow.hpp in your C++14 (or higher) project. To enable the built-in HTTP Client for secure cloud provisioning, link libcurl and define -DWILLOW_ENABLE_CURL.

#include "willow.hpp"

// 1. Provision Model (RAM-Only)
std::vector<uint8_t> buffer = fetch_from_cloud();
willow::Model model = willow::Model::load_from_memory(buffer);

// 2. Initialize Engine
willow::Detector detector(model);
willow::DetectionResult result;

// 3. Process Live Data
willow::Skeleton frame = get_76_point_skeleton();
if (detector.push_frame(frame, result)) {
    std::cout << "Action Detected! Confidence: " << result.confidence << "\n";
}

---

📚 Documentation

Please refer to the docs/ folder for comprehensive architectural guidelines:

1. Architecture & Best Practices - Understand how the Willow system functions.
2. Topology & Transforms - The Willow 76-point joint index reference & ROS/Unity bridging. 3.Provisioning & DRM - How to fetch, cache, and manage secure .int8 models via libcurl.
3. Evaluator Physics - Scoring speed, smoothness, and impulse at the edge.
4. Retargeter Kinematics - Extracting Quaternions for RL & Simulation.

🛠️ Building Examples & Tests

# Run the Golden Vector Parity Test (Verifies V40/V41 parsing)
g++ -std=c++14 tests/test_golden_vector.cpp -o test_willow
./test_willow

# Run Examples
g++ -std=c++14 examples/01_basic_detection.cpp -o ex01
./ex01

Support & Licensing

The Willow Runtime is a premium commercial service. A Partner License is required to provision models.

• Request a License
• Technical Support

© 2026 Willow Dynamics. All rights reserved.