| name | yahboom-robots-expert |
| description | Use this skill when you need domain expertise on Yahboom Raspbot v2 hardware, ROS 2 setup, mecanum kinematics, or integration architecture. |
Yahboom Robots Expert Skill
Use this skill when you need domain expertise on Yahboom Raspbot v2 hardware, ROS 2 setup, mecanum kinematics, or integration architecture—beyond routine operator procedures (see yahboom-operator for day-to-day tool use).
🛠️ Hardware Specifications
- Platform: Yahboom Raspbot v2. Main compute: Raspberry Pi 5 (ARM64). OS: Ubuntu 24.04 + ROS 2 Humble.
- Drive: Four mecanum wheels (32.5 mm radius). Layout: front_left, front_right, back_left, back_right; axles parallel to robot X (forward). Enables holonomic motion: forward/back, strafe left/right, rotate in place.
- Sensors: 2D LIDAR (laser_link), camera (camera_link); IMU and wheel encoders for odometry.
- Power: Onboard battery; monitor via
read_battery / telemetry. Low battery (< 20%) limits recommended motion duration.
📐 Frames and URDF
- base_footprint: Ground plane (z = 0).
- base_link: Body frame; z offset = +wheel_radius (0.0325 m) above ground. Wheel joints at z = -wheel_radius relative to base_link.
- Wheel positions (joint xyz): x = ±0.08, y = ±0.0745 (half track), z = -0.0325.
- LIDAR: Offset (0, 0, 0.0825) from base_link.
- Camera: Offset (0.105, 0, 0.05).
🤖 ROS 2 Architecture
- ROS 2: Humble on Ubuntu 24.04. Topics/services follow ROS 2 naming (e.g. cmd_vel, odom, scan, image).
- MCP ↔ Robot: yahboom-mcp server talks to a REST bridge (or direct ROS 2 node) on the robot network. Robot IP and bridge port configurable (env or settings).
- Dual mode: Server can run
--mode stdio (MCP only) or --mode dual --port 10892 (REST API + MCP). Dashboard (Vite) uses REST on 10892; frontend dev server on 10893.
⚙️ Kinematics and Motion
- Mecanum rollers: rollers at 45°; combined wheel velocities yield body linear and angular velocity.
- Odometry: from encoders + IMU; heading from IMU.
- Trajectory recording: start_recording → run motions → stop_recording(basename).
🌐 Integration and Fleet
- Federated fleet: Yahboom-MCP is one node; others include Dreame-MCP (mapping/sweeping), Virtual-Robotics-MCP (simulation), central hub (orchestration).
- 3D viz: Dashboard loads real STL meshes from URDF (base_link, wheels, LIDAR, camera).
⚖️ Skill Application Matrix
| Use yahboom-robots-expert when | Use yahboom-operator when |
|---|
| Explaining hardware, frames, or kinematics | Running tools and following workflows |
| Debugging connection, URDF, or frame issues | Choosing which tool/prompt to call |
| Designing integrations or fleet behaviour | Doing health_check → motion → recording |
| Answering “how does the robot work?” | Answering “how do I make it patrol?” |
🔗 Quick References
- Tools: yahboom, yahboom_help, yahboom_agentic_workflow.
- Prompts: yahboom_quick_start(robot_ip), yahboom_patrol(duration_seconds), yahboom_diagnostics().
- Scripts:
scripts/check_health.py (REST health/telemetry), scripts/run_patrol_square.ps1 (run server + dashboard hint).
- Docs:
docs/CONNECTIVITY.md, docs/architecture.md, docs/fleet_overview.md.
