Tabletop HandyBot

A low-cost AI powered robotic arm assistant that listens to your voice commands and can carry out a variety of tabletop tasks. Currently it can pick and place arbitrary objects presented in front of it. The total BOM is around $2300 USD.

Video Demo

System Architecture

Technology Used

• AR4 Robot: a $2000 robot arm that can lift 5lb dumbbells
  • ar4_ros_driver: ROS driver for the arm
• Realsense D435: a reliable $300 RGBD camera
  • realsense-ros: ROS wrapper for the camera
• Whisper Mic: Convenient library for using OpenAI Whisper with a microphone
• Grounding DINO A SOTA zero-shot object detector that can detect any object.
• Segment Anything: A SOTA zero-shot object segmentation model that can segment any object.
• OpenAI Assistants: an API that calls ChatGPT with specific instructions and function calls

Setup

Pre-requisites

You should have ROS 2 Iron installed on Ubuntu 22.04.

The follow hardware is is used. The use of other hardware will require adapting the codebase.

• AR4 robot arm with AR servo gripper
• Intel RealSense D435 Depth Camera

You should be able to run ar4_ros_driver with the gripper, and successfully perform hand-eye calibration.

Install

Import dependent repos

vcs import . --input tabletop-handybot.repos

Create a virtual environment, i.e.

pyenv virtualenv 3.10.12 handybot

Go to the ./Grounded-Segment-Anything/Grounded-Segment-Anything/ sub-directory and setup Grounded-Segment-Anything in the virtualenv. Ensure you can run the grounded_sam.ipynb notebook.

Install Python dependencies not in the ROS Index

pip install -r requirements.txt

Build and source the project

colcon build
source install/setup.bash

Run

Launch the AR4

ros2 launch ar_hardware_interface ar_hardware.launch.py calibrate:=True include_gripper:=True

Launch all other programs

ros2 launch tabletop_handybot run.launch.py

If all things are looking good, then you can try publishing a prompt to the /prompt topic for the robot to execute, i.e:

ros2 topic pub --once /prompt std_msgs/msg/String "data: 'put the marker in the container'"

If you have a microphone attached to the computer, you can publish a message to the /listen topic, and then say your prompt.

ros2 topic pub --once /listen std_msgs/msg/Empty "{}"