T-Top

T-Top is a tabletop robot with advanced audio and vision sensors, deep learning perception and telecommunication capabilities for richer interaction with people.

Home
Works with AI
ROS
3D printable

Suggested by Robotheca, created by @introlab (opens in a new tab), from github.com/introlab/t-top (opens in a new tab). Pinned to commit 8d1b471 (opens in a new tab).

Fork Download

Hardware license: CC-BY-NC-SA-4.0 (opens in a new tab)
Software license: GPL-3.0-only (opens in a new tab)
Commercial use: Not allowed

A non-commercial license (CC-BY-NC) does not allow selling kits or parts.

This repository contains executable files that Robotheca doesn't check. Only run them if you trust the creator.

Which files (6 scripts)

Its license doesn't let Robotheca show its photos.

See the photos on GitHub

The 3D files are too large to show here. View the original files from GitHub (185.2 MB) (opens in a new tab)

What it can do

Written by Robotheca from the project's documents.

Tabletop robot with a rotating base and a Stewart platform head for motion.
Audio perception from a 16-microphone array with sound localization and speaker identification.
Vision from an RGB-D camera and a wide angle camera, supporting face and person perception.
Perception and behaviors include SLAM, object detection, speech to text, face following, and telepresence.
Communicates with people through a touchscreen, LEDs, speakers and a battery status display.

Works with AI

Found by Robotheca in the project's files at commit 8d1b471, its robot.yaml and Robotheca's notes on its documents. Never guessed; Robotheca doesn't run the code.

Ships trained model files: en_US-amy-low.onnx and 3 more
In the project files: ros/behaviors/piper_ros/models/en_US-amy-low.onnx
Uses computer vision: YOLO
In the project files: ros/tests/yolo_performance
Uses speech recognition: Whisper
In the project files: ros/perceptions/speech_to_text/launch/test_whisper_speech_to_text_node.launch.xml
Uses a machine-learning library: PyTorch
In the project files: tools/dnn_training/object_detection/modules/convert_darknet_cfg_to_pytorch_module.py
Uses a machine-learning library: ONNX
In the project files: ros/behaviors/piper_ros/models/en_US-amy-low.onnx.json

Bill of materials

This repository has no parts list Robotheca can read yet.

Build guide

The guide appears here as you scroll.

Guide by @introlab (README.md). Its license (CC-BY-NC-SA-4.0 (hardware) and GPL-3.0-only (software)) doesn't let everyone reuse it. To reuse it, ask the creator on GitHub (opens in a new tab). Creators can ask Robotheca to remove it at contact@robotheca.org.

More guides

Wiring and electronics

Notes written by Robotheca from the project's documents.

Main parts

NVIDIA Jetson AGX Xavier or AGX Orin developer kit as the main computer.
Teensy 4.0 as the motor microcontroller.
Teensy LC as the battery microcontroller.
Dynamixel XL430-W250 servo motors for the base and the Stewart platform.
16 microphones with a USB sound card, Intel RealSense D435i and an Arducam wide angle camera.
Power adapter at 19 V, a rechargeable battery pack with a charger, and an INA220 current/voltage sensor.
Audio amplifiers, speakers, cooling fans, a 7 inch touchscreen, and an Intel WiFi card.

Connections

Custom PCBs for the buck-boost converter, Dynamixel control and power/battery control are documented in the repository.
Power enters through a 19 V adapter and is regulated by a buck-boost board before reaching the electronics.
The battery management is handled by its own microcontroller board on the power side.

Schematics, boards and diagrams

Project information

Difficulty: Advanced
Estimated cost: Not stated
Buildability: 69/100Buildable with some effort
Parts: –
Tools: 3D printer

Buildability69/100

Buildable with some effort

How complete the published files are for rebuilding this robot, measured by Robotheca at commit 8d1b471. It isn't a rating of the robot. Counted out of 80 points and scaled to 100: the parts that don't apply to this robot are left out. How we score

3D and CAD files20/20
32 printable parts (STL, 3MF), 121 CAD source files.
Robot descriptionNot counted
No URDF. Not counted: the score is made of the parts that apply.
Optional: add a URDF, and the robot can be seen in 3D and tried in the browser.
Parts list0/20
No parts list in the repository or linked.
Most points to gain: Link to your parts list, or add it to the repository (a CSV or a Markdown table) with quantities, prices and where to buy each part.
Software15/15
Code in this repository, instructions to install or run it, built on ros.
Documentation10/15
README of 4,613 characters with 3 pictures; 0 build steps.
Split the build into at least 3 steps.
SimulationNot counted
No simulation files. Not counted: the score is made of the parts that apply.
Optional: add a simulation model (MuJoCo, Gazebo, Webots or Isaac).
Licenses10/10
Hardware: CC-BY-NC-SA-4.0; software: GPL-3.0-only.

Software

Framework: ROS
Languages: Python, C++, C, JavaScript, Shell, Arduino
License: GPL-3.0-only

Software that runs on this robot

No software from Robotheca's catalogue found in this robot's files yet. Links come from its robot.yaml, its dependency files read as text, or Robotheca's editors.

The code

How to run it

Written by Robotheca from the project's documents.

Set up the development computer as described in the computer configuration documentation for Ubuntu 20.04.
Flash the firmware for the Teensy microcontrollers from the firmwares folder.
Build and run the ROS packages in the ros folder on the Jetson computer.

Robotheca doesn't review code. Read it before running it.

Project files

Every file of the project: code, CAD, electronics, guides, at commit 8d1b471.

Community

Questions and improvements go to the creator's repository.

Photo and 3D model