Introduction

Roadmap

  • Brand identity & visuals (Rolled out)
  • Cinematic brand reveal (Launched)
  • Website & documentation (Public-ready)
  • Social channels activated (Aligned & live)
  • Launch execution plan (Early engagement)

Goal: Prepare ecosystem for launch.

Vision API (VIBR)

  • Object and face detection
  • Multi-object tracking with track IDs
  • Behavior/event analysis (aggression, loitering, etc.)
  • Outputs mapped to vision_msgs/Detection2DArray

Voice API (VCCM)

  • Offline STT (streaming + batch)
  • Intent parsing with slot filling
  • Outputs mapped to std_msgs/String and qace_msgs/Intent

Qognix Runtime API

  • Rule-based decision logic via JSON/YAML rules
  • Supports conditional triggers, temporal logic, and multi-source inputs
  • Outputs actions as qace_msgs/Action or MQTT topics

ROS2 Bridge Package

  • Container + ament package
  • Subscribes to QACE API streams and republishes as ROS2 topics
  • /vibr/detections, /vibr/tracks, /vccm/transcript, /vccm/intent, /qognix/actions

Navigation API (SLAMGrid – SGNS)

  • Real-time 2D/3D mapping with RTAB-Map or Cartographer
  • Robot localization within generated maps
  • Anchors detections from VIBR into SLAM space for spatial awareness
  • Outputs mapped to nav_msgs/OccupancyGrid and geometry_msgs/PoseStamped

Control Output API (QACE Bridge)

  • Converts abstract Qognix actions into robot-consumable commands
  • Supports ROS2 publishers and MQTT messaging for distributed robots
  • Example: “goto:kitchen” → navigation goal published as PoseStamped

Unified AI Layer (Qognix Expansion)

  • Orchestration of vision, voice, navigation, and control through a single API surface
  • Multi-modal event fusion (e.g., “If person=Alex AND zone=kitchen → send robot”)
  • Streamlined behavior definition with JSON/YAML rules

Extended ROS2 Bridge

  • Adds new topics for navigation and control outputs
  • /sgns/map, /sgns/pose, /bridge/cmd plus existing VIBR/VCCM/Qognix streams
  • Includes diagnostics and heartbeat for system monitoring

QACE SDK (Alpha Release)

  • ROS2-compatible packages for all QACE modules (VIBR, VCCM, Qognix, SGNS, Bridge)
  • Python and C++ client libraries for direct integration
  • Offline execution on Jetson, x86, and ARM boards

Local Runtime Support

  • Runs APIs as local containers or libraries without external connectivity
  • Optimized for 5–10 Hz event loops, with higher rates depending on workload and hardware
  • Inference accelerated through TensorRT and ONNX Runtime

Developer Tools

  • Prebuilt launch files for ROS2 Humble and Foxy
  • CLI utilities for module orchestration, monitoring, and logging
  • Simulation-ready packages for Gazebo and Isaac Sim

Extended Integration

  • Multi-camera pipeline support and sensor fusion
  • Synchronization of VIBR (vision) and SGNS (SLAM) for context-aware navigation
  • Output signals mapped to qace_msgs custom message types for intent, action, and events

QARC Hardware

  • Compact edge computer preloaded with the full QACE stack (VIBR, VCCM, Qognix, SGNS, Bridge)
  • Hardware options:
    • Lite: Jetson Nano/Orin Nano for education and prototyping
    • Pro: Jetson Xavier NX/Orin NX for full robotics stack deployment
    • Hybrid: x86 with RTX-class GPU for industrial robotics and multi-robot coordination
  • Wide I/O support: 4× GigE (2× PoE+), USB, GPIO, CAN, RS-232/485, and optional cellular

Preinstalled Software Stack

  • Full QACE runtime (APIs + Unified Layer + SDK) pre-configured
  • ROS2 (Humble/Foxy) and Docker support out of the box
  • Local dashboard for monitoring modules and orchestration
  • Update toolchain for module upgrades and patches

On-Device Execution

  • Runs all QACE modules locally with predictable performance
  • Inference accelerated with TensorRT and ONNX Runtime
  • Supports real-time event loops for perception, navigation, and control

Deployment Ready

  • Fanless industrial-grade design with extended temperature tolerance
  • Ruggedized option for outdoor, logistics, and defense robots
  • Pre-validated integrations with selected sensors (Intel RealSense, OAK-D, RPLIDAR) and actuators via QACE Bridge
Previous

Key Features

Next

Architecture