All user builds

User build

AR4-Powered AI Assistant for Voice-Controlled Pick

Expert/ProfessionalRobotics Engineering

Listens to voice commands to autonomously pick/place objects using AI

Built by Yifei ChengToronto, Ontario, Canada

1

Photo

AR4-Powered AI Assistant for Voice-Controlled Pick

About my build

This AR4-based “HandyBot” understands voice commands to autonomously pick/place tabletop objects using cutting-edge AI vision (Grounding DINO, Segment Anything) and OpenAI Whisper speech recognition. The $2300 system combines an AR4 arm, RealSense D435 camera, and custom ROS 2 pipeline for real-world object interaction.

🔗 Build it yourself: GitHub Repo
💬 Discuss the ROS driver: Community Thread

Workflow / process

  1. Voice Command: User speaks a prompt (e.g., “Put the marker in the container”).

  2. AI Processing: OpenAI Whisper transcribes speech, Grounding DINO detects objects, and Segment Anything creates masks.

  3. Action Execution: AR4 ROS driver calculates grasps and moves the arm to complete the task.

Primary use cases

  • AI robotics research
  • Object sorting system
  • Voice-controlled automation

Custom modifications

  • Custom mic setup
  • Hand-eye calibration
  • ROS 2 integration

Hardware add-ons

  • RealSense D435
  • Servo gripper
  • Whisper mic setup

Software used

  • ChatGPT API layer
  • Grounding DINO
  • OpenAI Whisper

Comments

No comments yet.

Community

Other AR4 robot builds