Autodiscovery HEAD-lite (Version 3)

Autodiscovery HEAD-lite

The HEAD-lite Version 3 is the lightweight, outdoor-optimized counterpart to the HEAD360. Designed for compact mobile robots in field environments like agriculture and mining, it integrates high-precision GNSS/GPS for outdoor localization, one (1) wide-FOV 4D depth camera with standalone NPU edge processing, and high-capacity solid-state storage (using the Autodiscovery Vault drive system) for long-duration offline deployments.

Technical Specifications

Sensors & Vision

Feature Specification
Primary Navigation Sensor Integrated high-precision GNSS/GPS receiver for outdoor localization and mapping.
Vision & Depth One (1) wide-FOV 4D depth camera. Support for optional external 3D LIDAR or vision payloads.
Night Operation Built-in high-power LED flood lights for illumination in zero-lux environments.

Computing, AI & Storage

Feature Specification
Edge AI Processor Dedicated on-board Neural Processing Unit integrated into the 4D camera for standalone app execution.
Main Compute Node (High-Performance Tier) Advanced AI compute node with 128GB memory for intensive neural network workloads.
Main Compute Node (Power-Efficient Tier) Energy-efficient x86 compute node with 32GB RAM for sustained deployment.
Data Storage High-capacity solid-state storage (Autodiscovery Vault drive system) for high-bandwidth data logging during field trials.

Networking & Communications

Feature Specification
Router Built-in advanced multi-gigabit Wi-Fi 7 network routing device.
Cellular 5G communication support via optional USB expansion.
Audio Integrated multi-microphone array and speakers for voice.

Mechanical & Electrical

Feature Specification
Power Supply Built-in power regulation supporting a wide input range of 24V – 90V DC.
Environmental IP65 rated from above (fully IP65 rated when integrated and sealed correctly on the host robot).
Thermal Integrated active and passive thermal management system.

Software & Compatibility

Feature Specification
Software Stack Custom open-source software built for ROS 2 Jazzy and Navigation 2 (Nav2). Accessible to customers via the Autodiscovery GitHub.
Middleware Zenoh DDS for low-latency network communication.
Platform Compatibility Out-of-the-box compatibility with a wide range of quadruped, wheeled, and tracked robotic platforms including Unitree A2 and A2-W, Unitree B2, AgileX robots (Scout Mini, Bunker Mini, Ranger, Bunker Pro), and many more.
Autodiscovery HEAD-lite Details

Core Capabilities Explained

GNSS Mapping and Autonomous Navigation

The HEAD-lite leverages its integrated high-precision GNSS receiver combined with spatial depth data from its 4D camera to perform reliable outdoor localization and mapping. The pre-configured Navigation 2 (Nav2) stack uses these real-time data inputs to execute safe waypoint navigation, dynamic path planning, and robust obstacle avoidance, making it ideal for agricultural and surveying applications.

Edge AI Processing & Long-Duration Storage

To maximize efficiency and robustness, the camera module features an independent NPU co-processor. This allows the HEAD-lite to run lightweight object detection models natively on the edge without using central system resources. For extensive research runs, the high-speed Vault SSD storage acts as a high-bandwidth data-logging repository, capturing high-definition video, metadata, and sensor files offline in the field.

Advanced Detection Capabilities

With its edge-to-core compute architecture, the HEAD-lite natively runs multiple pre-trained machine learning models:

  • Human & Demographic Analysis: Detects people, tracks multi-person movement, performs facial recognition, and extracts demographic and sentiment data.
  • Security & Logistics: Runs Automatic License Plate Recognition (ALPR) to identify and log vehicles in real-time.
  • Custom Models: Supports custom-trained neural networks for specific field tasks (like plant disease classification, weed detection, or anomaly detection).

Voice Control and Voice Interaction

Equipped with high-fidelity speakers and a multi-microphone array, the HEAD-lite supports natural language processing (NLP) to interpret operators' verbal instructions. It can also speak to report status updates, warn bystanders, or act as an active intercom node for remote field operators.

ROS 2 Ecosystem and Zenoh DDS Advantages

The HEAD-lite is built entirely around the ROS 2 Jazzy ecosystem, ensuring seamless compatibility with standard packages, simulators, and existing robotics codebases. The underlying Zenoh DDS middleware provides low-overhead, low-latency network communication, keeping the control and telemetry streams stable even over unstable, long-range Wi-Fi or 5G cell connections in isolated field trials.