The Anthro (Anthropomorphic Robot)

Any-tasks, fully modular humanoid robotics platform.

Similar projects worth following
The Anthro (or Anthropomorphic Robot) is the robotic platform we are currently developing, and has been in development since 2015 (which began as Project '87). The platform is intended to be fully modular, lightweight, and endlessly customizable. The assembly of the Anthro is 80% toolless, making production quick, and customizing easy for the end user.

Through a combination of autonomous cognitive and human-assisted training, the Anthro will be capable of performing many tasks. We intend to make the Anthro the most affordable, and most customizable humanoid robot available. The Anthro is driven by brushless DC (BLDC) motors controlled by a number of different BLDC drivers (including ODrive and in-house custom boards) to facilitate fast, high torque, closed-loop control, with added safety. High power BLDC motors and zero-backlash reduction systems provide the Anthro with enough power to perform walking, running, jumping and heavy lifting.

What is the Anthro?

The Anthropomorphic Robot is our answer to the lack of existing full-size humanoid robots available to the general public. What is preventing people from owning their own right now? There are three factors:

  1. Cost: Even if there were one available right now, it would cost far too much for the average person to own and maintain.
  2. Availability: Apart from Atlas and Digit, there aren't too many humanoid robots on the market, let alone available to the general public.
  3. Ease of Use: Research platforms and industrial robots are great for professionals and developers, but aren't suitable for everyone. 

Other issues that could arise from other robotic platforms include:

  • Closed Source Hardware Design
  • Privacy and Data Protection
  • Offline and Online Operation
  • Customization and Personalization

We aim to solve these issues by developing the Anthro with both the consumer and professional in mind. Here's how we are doing this:

  • Develop the Anthro using as many off-the-shelf components as possible. Use 3D-printed parts wherever possible to reduce cost and weight. 
  • Make the hardware design open-source where applicable. This is to enable third-party developers to make improvements to components. This also allows further customization by the end user.
  • Make the design semi-toolless for easy assembly, repair and modification. Training and interacting with the robot will be intuitive and seamless - like talking to another person.
  • Harness the power of cloud computing whenever possible. An Anthro can utilize our GPU servers to accelerate visual inference, speech processing, and training. 
  • Have a respectable amount of onboard computing power. The robot should be capable of operating on its own whenever internet connection is not available.
  • Every Anthro should be unique! Users can customize each one's appearance: from material to color scheme. One could also have their Anthros be uniform in appearance for commercial applications.

What will you find on this project page?

This page will focus primarily on the hardware side of the platform. A separate page may be created later for the software/AI side of things. Project logs will start from the beginning of the project (circa 2014), and work their way to where we are currently at, to better show how this project has evolved over the years. 

  • First venture into Brushless actuators

    Anthrobotics02/12/2021 at 15:16 0 comments

    In this update, we look into our first attempts at using brushless motors paired with 3D-printed gearboxes to create powerful backdriveable actuators. The first ones were based on the Opentorque actuator made by @Gabrael Levine . The motors are controlled using the ODrive motor controller.

  • Modular Foot Design and Object Detection

    Anthrobotics02/04/2021 at 05:07 0 comments

    In this week's update, we will take a look at our first 3D-printed modular foot design, as well as an object recognition demo. We run YOLOV2 on the Nvidia Jetson TX2 to detect different types of fruit. A side-by-side arrangement of webcams served as a stereo camera, using dual video streams to eventually be used for depth perception and increased simultaneous detections.

    Stay tuned for more updates next week, where we will take a look at our first 3D-printed brushless actuator! #robotics #3DP #stereo #AI #anthro

  • Speech Recognition Hardware Demo

    Anthrobotics01/28/2021 at 18:03 0 comments

    In this video, we tested out and compared two different speech recognition systems. One of them is based on a microcontroller solution, MOVI, while the other is based on Pocketsphinx running on the Raspberry Pi.

  • Testing new arm actuators

    Anthrobotics01/24/2021 at 06:42 0 comments

    In the following videos, we tested out two new arm actuation methods. The first uses servo motors assisted by helical 'artificial muscles', with the second using stepper motors. This was our first venture into testing robot designs powered by stepper motors.

  • Testing (Faux) Fur as a "skin" option

    Anthrobotics01/22/2021 at 04:14 0 comments

    In this update video, we demonstrated the use of faux fur as both an aesthetic and functional covering for the Anthro. We also demonstrate the compatibility of Force Sensing Resistors (FSRs) with fur to act as a simple type of "robot nerve".

  • Project 87 Update - January 12, 2016

    Anthrobotics12/30/2020 at 03:48 0 comments

    "Hey everyone. I've been working tirelessly to get this video up for you guys to see what we have accomplished so far. Thanks for watching, and expect more updates often!" 

    - @Fennex628 

  • Servo Arm Drives in Action

    Anthrobotics12/30/2020 at 03:46 0 comments

    Here are two demo videos of the servo-driven arms on the robot from 2016:

  • Geared Servo Arm Drives

    Anthrobotics12/30/2020 at 03:44 0 comments

    A set of arms now accompanies the torso!

    Both arms are made of aluminum channel, and are also driven using hobby servos like the remainder of the robot. A simple linear gear drive using plastic gears is used to drive the elbows. A single large servo powers the shoulder, and wrist rotation servos are mounted in the forearms.

    Finally, Freddy now sports a sweet pair of running shoes :)

  • Endoskeleton Design Improvements

    Anthrobotics11/25/2020 at 05:48 0 comments

    Between July and November 2015, a number of improvements were made to the endoskeleton!

    First off, that weird "double knee" thing going on was removed, and a single-axis design was implemented using a servo motor on each side.

    Next, we made the foot smaller, and switched to aluminum from steel. A hinged front toe was also added.

    Next up, arms would be added to the torso.

  • The first steps!

    Anthrobotics11/25/2020 at 05:34 0 comments

    We had hit our first milestone: getting our robot to take it's first steps!

    Of course, it wasn't entirely unassisted. The legs lacked some essential degrees of freedom required to correct itself if it were to lean too far to either side. It also lacked an IMU of some kind, so it wouldn't have any idea as to its orientation with the environment. Hence @Fennex628 being there to provide a helping hand.

    At this point we had achieved very basic lateral motion moving in one direction, but it was pretty good for both our budget and timeframe at the time. Next, would be to add the additional DoF needed for advanced movement, and continue with the remainder of the build.

View all 16 project logs

Enjoy this project?



brucejdii wrote 02/12/2021 at 23:55 point

Push this concept onto Halloween Companies!  It’ll be a slam dunk billion dollar day - every Oct 31st!  Just to ballpark – how much would the “Erector Set” (frames and gears) set you back {excluding electronics}?

  Are you sure? yes | no

Anthrobotics wrote 02/13/2021 at 00:26 point

Haha thats a great idea! 😄

The aluminum/steel frames and gears cost us between $1700 and $2300 (Canadian) about 4 years ago. We expect the new 3d printed parts and carbon fiber frame to keep the cost much lower! Electronics are always the more expensive part for sure. 

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates