The Operating System
for BonicBot
BonicOS is the high-performance onboard software layer powering every BonicBot. It bridges complex physical actuators and edge sensor arrays with real-time AI and cloud coordination.
$ bonicos status --fleet[INFO] BonicOS connection: ESTABLISHED to bonic.ai[INFO] Actuator controller: OK (16 DOF calibrated)[INFO] Onboard LIDAR: 360° SCANNING (10Hz)[INFO] Local Edge LLM: LOADED (Quantized 7B parameters)[INFO] Current battery: 84% (Charge cycle 42)$ python -c "import bonic_os"# SDK interface ready. Listening on localhost:50051
Empowering Robots with Embodied Intelligence
BonicOS runs natively on the onboard computer of every BonicBot. By compiling high-speed motor kinematics, vision pipelines, and sensor arrays into a standard, accessible API, it abstracts away low-level complexities.
Whether navigating classroom corridors, greeting guests at reception halls, or conducting inspections on warehouse floors, BonicOS guarantees safety, reliability, and immediate response.
Every BonicOS deployment connects seamlessly to the BonicAI platform, where the operating system and robot behaviors are configured. From BonicAI, you can set up autonomous agents, collect data streams directly from the OS, perform remote teleoperation, and run real-time hardware diagnostics through a unified management layer.
The Bonic Software Stack
How BonicOS acts as the middle-tier intelligence layer, connecting physical robot bodies with cloud intelligence.
04 | End Users
Teachers, operators, enterprise administrators, and researchers who orchestrate workflows.
03 | BonicAI Cloud Platform
Central fleet dashboard, AI lesson planners, marketing studios, remote teleoperation pipelines.
02 | BonicOS (Onboard Engine)
The critical intelligence layer executing local sensor fusion, SLAM navigation, edge AI models, and ROS2 routing.
01 | BonicBot Hardware
Physical actuators, motor controllers, depth sensors, LIDARs, and computing blocks (A, S, and M Series).
BonicOS Onboard Intelligence
BonicOS operates across multiple physical layers, bridging real-time microcontroller firmware and central CPU processing with edge AI models and interactive interfaces.
Bare-Metal Firmware
Real-time Control Board
Runs directly on onboard microcontrollers. Powers sub-millisecond real-time motor PID loops, joint torque calibrations, hardware telemetry feedback, and deterministic safety limiters.
ROS2 Processor Core
Central CPU (ARM/x86)
The OS core built natively on ROS2. Manages node communications, tf2 coordinate frames, LiDAR point-cloud projection, SLAM mapping, and low-level peripheral drivers.
Edge & Cloud AI Engine
Central Processor NPU/GPU
Houses the primary Vision-Language-Action (VLA) models, edge LLM text/reasoning pipelines, offline edge Speech-to-Text (STT) and Text-to-Speech (TTS), and WebSocket telemetry sync.
Display Interface
Onboard Interactive Display & Phone Screens
Runs on the physical robot display or connected phone. Used for presenting dynamic content generated or assigned from the BonicAI platform, collecting customer data and feedback forms, and showing interactive information.
Core Capabilities
Engineered with advanced components that allow BonicBots to interact, learn, and act safely.
ROS2 Native Integration
Built directly on ROS2 . Native support for publishers, subscribers, action servers, and tf2 transforms. Leverage the industry standard.
On-Device AI Inference
Run edge Large Language Models (LLMs) locally on-device. Local Text-to-Speech (TTS) and Speech-to-Text (STT) ensure low latency and offline functionality.
Computer Vision Stack
Real-time visual processing: YOLO-based object detection, face recognition, spatial estimation, and visual SLAM mapping.
Real-Time Sensor Fusion
Unifies data feeds from 360° LIDAR, IMU, depth cameras, and ultrasonic arrays into a single, cohesive spatial occupancy grid.
Hardware Safety Layer
Deterministic, low-level collision avoidance. Instant motor braking and sensor-level fault monitoring run continuously on-board.
Secure OTA Updates
Deploy security patches, motor calibrations, and AI model updates from BonicAI directly to the edge with cryptographically signed bundles.
BonicAI Cloud Bridge
Synchronises live robot telemetry, coordinate mappings, and fleet statuses with BonicAI platform via standard WebSockets.
VLA Task Executor
Supports advanced Vision-Language-Action (VLA) models, translating natural voice directives into multi-step physical manipulations.
Supported Platforms
BonicOS is optimized for deployment on all three series of the BonicBot family.
BonicBot A Series
Layer 4 // Interface
Android Phone Interface Layer
Layer 3 // Intelligence
Runs on both the Processor and Phone
Layer 2 // Onboard CPU
Raspberry Pi 4 (Pro Models only)
Layer 1 // Controller Board
Custom Controller Board (Firmware runs here)
BonicBot S Series
Layer 4 // Interface
Android-based Onboard Interactive Display (attached to chest)
Layer 3 // Intelligence
Runs on both the Processor and Onboard Interactive Display
Layer 2 // Onboard CPU
Nvidia Jetson Nano or Raspberry Pi (Pro Models only)
Layer 1 // Controller Board
Custom Controller Board (Firmware runs here)
BonicBot M Series
Layer 4 // Interface
Android-based Processor Interactive Display (on robot)
Layer 3 // Intelligence
Runs on both the Phone and Onboard Interactive Display (advanced)
Layer 2 // Onboard CPU
Nvidia Jetson Nano / Orin NX / Nvidia Thor (Pro Models only)
Layer 1 // Controller Board
Custom Controller Board (Firmware runs here)
A Developer-Friendly Platform
BonicOS is designed with extensibility in mind. Built on a fully integrated ROS2 structure, researchers and AI developers can directly interact with the robot, customize behaviors, and interface with motor controllers.
Connect custom software packages, train reinforcement learning policies, or fine-tune vision models. Our Python SDK communicates securely with the robot to enable advanced control.
- Full Python SDK support
- Secure API endpoints via BonicAI API keys
- Custom motor tf2 transformations
import rospy
from bonic_os import BonicBot
class RobotManager:
def __init__(self):
# Initialize onboard BonicOS client
self.bot = BonicBot(series="S Series")
def start_interaction(self):
# Trigger natural language conversation
self.bot.ai.speak("Hello! How can I assist you today?")
# Subscribe to spatial detection events
self.bot.vision.on_person_detected(self.greet_person)
def greet_person(self, person_id, details):
print(f"Recognized guest: {details.name}")
self.bot.navigation.move_to(x=1.2, y=0.5)
self.bot.ai.speak(f"Welcome back, {details.name}!")Build on BonicOS
Get the SDK, explore motor calibrations, or request custom integrations for your school, lab, or company.
Get in Touch