crusbro robot-agent for Intelligent Robots

ROBOT · LLM + SKILLS + TOOLS · SAFE ORCHESTRATION

The intelligent core for robots in the embodied-AI era. As embodied intelligence moves from proof of concept to industrialisation, competition is shifting from mechanical structure and single performance metrics to the completeness of the intelligence stack. Built on the crusbro-agent SDK, this product sits on top of OS + ROS2, providing a perception–decision–execution loop with continuous iteration—turning natural-language goals into executable, auditable and recoverable workflows.

Application business · UI crusbro robot-agent Capability · Action · Dialogue · Behaviour Sense → Decide → Act · Continuous iteration OS + ROS2 Topic · Service · Action HAL · Capability chassis · camera · LiDAR · arm
Intelligence stack · on top of OS + ROS2

Intelligent core, redefining what intelligence means on a robot

The crusbro robot-agent sits on top of OS + ROS2 as the intelligent core of next-generation robots. It intelligently orchestrates capabilities, actions, dialogue, behaviour and emotional expression—intelligence made observable.

It fuses cloud model capabilities with on-device real-time perception and mechanical execution into a sense → decide → express → act → recover loop, moving robots from "able to move" to "able to understand intent, collaborate proactively, and serve continuously".

1
Capability orchestrationUniformly coordinates navigation, vision, speech, manipulation, arms and IMU so modules stop working in silos.
2
Action orchestrationBreaks long tasks into steps with automatic retry, compensation and rollback—action stays controllable between safety and efficiency.
3
Dialogue orchestrationMulti-turn intent, context, identity and prompts stay in lockstep with task execution, so "what it says" matches "what it does".
4
Behaviour & emotionBrings scene awareness to facial expression, tone, pacing and motion amplitude, so the robot feels tuned rather than scripted.
Intelligent Core Capability nav · vision · speech Action plan · retry · rollback Dialogue intent · context · flow Behaviour face · tone · pacing
Four orchestration dimensions · Capability, action, dialogue and behaviour unified so the robot stops working in silos

Industrial-grade embodied AI is judged by the intelligence stack

The way robots are evaluated is changing: from mechanical structure and single performance metrics to the completeness of the intelligence stack, with generality and transferability replacing heavy customisation as the core criteria.

closed loop · continuous
Complete intelligence stack · One core spans perception fusion, planning, capability invocation and recovery
Agent Inspect Service Logistics
Cross-task & cross-scenario transfer · Skills/Tools capture experience, HAL isolates hardware
Data flywheel · continues Iterate
Data accumulation & iteration · Task data feeds back into Skills and policies; smarter with use
Sense see · hear · feel Decide plan · pick Skills Act move · control · recover Express speak · face · pacing
Sense → Decide → Act → Express · Intelligence is not a single feature—it is one continuous loop across four stages

Why an agent, not a cloud remote control

The value of a robot product is not "what model sits in the backend" but "whether the device sees, thinks and moves well". As the on-device intelligent core, crusbro robot-agent fuses cloud LLMs, Skills, Tools and real-time capabilities to turn intent into observable, auditable, recoverable behaviour.

Intelligence is a closed loop across sense, think, express and act. crusbro robot-agent assembles that loop from cloud models and on-device real-time capabilities, so the robot truly shows intelligence.

Seven-layer architecture with clean boundaries

Application → crusbro robot-agent orchestration → OS + ROS2 → capability layer → HAL → driver layer → hardware. Business apps never touch drivers or call LLMs directly; the capability layer exposes testable semantic interfaces through C++/Python SDKs and ROS2 Topics, Services and Actions, while HAL isolates differences between chassis, cameras, LiDARs and arms.

Application
business systems · UI
crusbro robot-agent orchestration
planning · memory · sandbox · Skills/Tools
OS + ROS2
Topic · Service · Action
Capability + HAL + Drivers
chassis · camera · LiDAR · arm
Hardware
sensors · actuators

One orchestration model for robot applications

  • Interaction, Perception, Task, Safety
  • Identity, DeviceMgr, Nav, Content, Comm
  • Boot, OTA, LogAudit, Power
  • Apps submit tasks and context; the Agent picks Skills and calls authorised Tools

From goal to controlled action

  1. Understand the goal, identity and scene context
  2. Plan steps and verify permission, risk and device state
  3. Call navigation, vision, speech and arm capabilities
  4. Route critical actions through confirmation, approval or human takeover
  5. Persist the plan, tool calls, results and replay evidence

From "able to move" to "understands, collaborates, keeps serving"

StageWhat the intelligent core doesWhat users observe as intelligence
Sense and understandFuses multimodal input with identity, memory and scene context to form intent.The robot recognises you, follows context, and never asks the same question twice.
Think and planDecomposes goals into steps, picks Skills, and validates permission and risk.Tasks advance on their own, with the robot explaining its key choices.
Express and interactSelects expression strategies for dialogue, face, tone and motion amplitude.Communication has pacing; motion has restraint—not cold, not excessive.
Act and recoverDrives navigation, vision, manipulation and arms; compensates, rolls back or hands over on failure.Failures get an explanation, the action retries, and the robot never gets stuck.

One core, many robot forms

The same intelligent core transfers across inspection, service, companion and logistics robots, with task experience reused across scenarios.

Inspection
Plants, campuses, data rooms; anomaly detection and alerting.
Service
Greeting, reception, guidance, companion dialogue.
Logistics
Indoor / outdoor delivery, multi-floor, lifts and access control.
Companion
Home, elder care, long-term memory and emotional expression.

Cloud-edge collaboration with offline fallback

  • Cloud · LLMs, Skills, permission policy, device management, audit and business integration
  • Device · Keeps intelligence core: navigation, identity, anomaly response, task execution and safety control
  • Transport · MQTT for commands / events / state; HTTPS for media and configuration
  • Offline · Sense, decide, express and act still work; events replay by priority once reconnected
Cloud LLM · Skills · audit · business device mgmt · permissions Robot intelligent core · nav · identity anomaly · execution · safety MQTT commands / events / state · HTTPS media & config Offline still senses, decides, expresses and acts; events replay by priority
Cloud-edge · Cloud supplies the brain, the robot keeps the cerebellum; key tasks still complete offline
Identity & credentials X.509 / mTLS · JWT Topic ACL · HMAC anti-replay Data governance local-first · encrypted upload least privilege · HITL Governed · Auditable · Roll-back-safe signed OTA · A/B partitions maps & real-time control independent of cloud
Security & rollback · Identity, permission, audit and OTA upgrade in one closed loop

Security, privacy and rollback

  • Device identity · X.509/mTLS, JWT, Topic ACL and HMAC anti-replay
  • Data privacy · Identity and on-site sensitive data stay local first, encrypted upload per policy
  • High-risk actions · Least privilege, HITL and full audit
  • OTA & rollback · Signed OTA + A/B partitions; auto rollback on failure; maps and real-time control don't rely on cloud

Core value for robot products

Differentiated experience
Unified intent understanding and expression strategy keep "what it says" consistent with "what it does".
Governed autonomy
High-risk actions go through HITL; cross-vendor hardware isolated by HAL; the system keeps improving.
Shorter iteration
Skills/Tools plug in as modules, so new scenarios don't need to rewrite the stack.
Multi-line reuse
The same intelligent core powers inspection, service, companion and logistics robots.

Engineering delivery and acceptance

The integration prioritises proven chassis, depth cameras, LiDARs, mic arrays and arms, with HAL/SDK forming swappable vendor boundaries; every capability ships with fake / loopback implementations so CI runs without real hardware. Targets include cold start under 30s, interactive response under 2s, key anomaly detection under 1s, LAN MQTT latency under 200ms, and acceptance covers HIL stability, OTA drills, navigation traces, offline compensation and high-risk action auditing.

Intelligent robot solution View SDK Talk to us