askill
ugv-nanda-agent

ugv-nanda-agentSafety 90Repository

Turn Waveshare UGV robots into NANDA-compliant A2A agents with vision, navigation, and task capabilities.

0 stars
1.2k downloads
Updated 1/31/2026

Package Files

Loading files...
SKILL.md

UGV NANDA Agent

Turn your Waveshare UGV robot into a NANDA-compliant A2A agent that other agents can discover, command, and pay.

What It Does

┌──────────────────────────────────────────────────────────┐
│  UGV Robot (Waveshare)                                   │
│  ┌─────────┐  ┌─────────┐  ┌─────────┐  ┌─────────┐     │
│  │ Motors  │  │ Camera  │  │ Sensors │  │ Pan/Tilt│     │
│  └────┬────┘  └────┬────┘  └────┬────┘  └────┬────┘     │
│       └───────────┬┴───────────┴────────────┘           │
│                   ▼                                      │
│  ┌────────────────────────────────────────────────────┐ │
│  │  UGV NANDA Agent                                   │ │
│  │  - A2A protocol (/.well-known/agent.json)          │ │
│  │  - Skills: move, look, capture, sense              │ │
│  │  - Optional: X402 payments                         │ │
│  └────────────────────────────────────────────────────┘ │
│                   ▼                                      │
│  ┌────────────────────────────────────────────────────┐ │
│  │  Discoverable by other agents                      │ │
│  │  "Hey robot, patrol the warehouse"                 │ │
│  └────────────────────────────────────────────────────┘ │
└──────────────────────────────────────────────────────────┘

Supported Hardware

DeviceMotorsCameraArmCompute
WAVE ROVERESP32/RPi
UGV RoverRPi/Jetson
UGV BeastJetson
RaspRoverRPi

Quick Start

# On your Jetson/RPi with the UGV
git clone https://github.com/QUSD-ai/ugv-nanda-agent
cd ugv-nanda-agent
bun install
bun run start

Configuration

# config.yaml
ugv:
  url: http://localhost:5000  # UGV Flask server
  
agent:
  name: warehouse-rover-01
  port: 3010
  description: Autonomous patrol robot with camera

nanda:
  identity:
    did: did:key:z6Mk...  # Auto-generated if not set
  
payments:  # Optional X402
  enabled: false
  wallet: 0x...

Agent Skills

move_robot

{
  "skill": "move_robot",
  "input": { "leftSpeed": 50, "rightSpeed": 50 }
}

set_camera

{
  "skill": "set_camera",
  "input": { "pan": 45, "tilt": -10 }
}

capture_image

{
  "skill": "capture_image",
  "input": {}
}

Returns base64 JPEG image.

get_sensors

{
  "skill": "get_sensors",
  "input": {}
}

Returns battery, ultrasonic distances, IMU data.

patrol (compound skill)

{
  "skill": "patrol",
  "input": { 
    "waypoints": [[0, 0], [100, 0], [100, 100]],
    "captureAtWaypoints": true
  }
}

Multi-Agent Example

// Vision agent discovers and commands the UGV
import { NandaClient } from '@qusd/nanda-skill';

const client = new NandaClient();

// Discover robots on the network
const robots = await client.discover({ 
  capabilities: ['move_robot', 'capture_image'] 
});

// Command the first robot
const ugv = robots[0];
await ugv.call('move_robot', { leftSpeed: 30, rightSpeed: 30 });
await sleep(2000);
await ugv.call('stop_robot', {});
const image = await ugv.call('capture_image', {});

Architecture

┌─────────────┐     ┌─────────────┐     ┌─────────────┐
│ Claude/LLM  │────▶│ NANDA Agent │────▶│ UGV Robot   │
│ (planning)  │◀────│ (protocol)  │◀────│ (hardware)  │
└─────────────┘     └─────────────┘     └─────────────┘
                           │
                           ▼
                    ┌─────────────┐
                    │ Other Agents│
                    │ (discover)  │
                    └─────────────┘

Why NANDA + Robots?

  1. Discoverability — Robots announce themselves on the network
  2. Interoperability — Standard A2A protocol, any agent can command
  3. Composability — Vision agent + Nav agent + Arm agent = coordinated system
  4. Payments — Charge for robot services via X402 micropayments

Links

Install

Download ZIP
Requires askill CLI v1.0+

AI Quality Score

78/100Analyzed 3/28/2026

Well-structured technical reference for converting Waveshare UGV robots into NANDA-compliant A2A agents. Provides comprehensive hardware compatibility table, clear skill definitions with JSON examples, architecture diagrams, and a multi-agent TypeScript example. The metadata with category/hardware tags and emoji enhance discoverability. Limited only by its specific hardware focus (Waveshare UGV ecosystem), but the A2A protocol foundation makes it broadly applicable within that domain.

90
85
75
82
80

Metadata

Licenseunknown
Version0.1.0
Updated1/31/2026
PublisherQUSD-ai

Tags

apici-cdgithubllm