This repository contains the code for a robot that I am slowly working on.
This code is pre-release: you cannot yet build an Artie bot - not all of the software and hardware is ready!
The purpose of Artie is twofold: data collection and testing developmental robotics theories.
The vision is that Artie will be fully open source, 3D-printable, and as cheap as is feasible, while still being easy to use and extend.
Before you can use Artie, you need to build him.
Please note: you cannot yet build an Artie! These instructions will certainly change, the bill of materials will change, the schematics will change, etc.
Building Artie is composed of the following steps:
- Get your parts
- Flash the parts
- Build your bot
See here for the full instructions
The Artie ecosystem consists of the following:
- Artie: The actual robot itself - the PCBs, 3D printed parts, the single board computers (SBCs),
microcontroller units (MCUs), sensors, actuators, LCDs, etc.
- Firmware: The MCU firmware.
- Yocto Images: The custom embedded Linux images.
- Libraries: Libraries shared between all the software components.
- Drivers: Applications that run on the SBCs and interface with hardware.
- Artie CLI: The simplest way to control a physical Artie. Used mostly for testing.
- Artie Tool: A single tool to flash, test, build, release, etc.
- (Planned) Artie Workbench: A web app that allows an authenticated user to control and configure Artie.
- (Planned) Demo Stack: A demo that uses traditional robotics algorithms to control Artie.
- (Planned) Reference Stack: A reference implementation of developmental robotics theories used to control Artie.
- (Planned) Simulator: A simulated environment and simulated Artie for training and testing without a physical bot.
Artie is meant to be used to collect data in real-life social situations as well as to test theories of developmental robotics.
There are three planned ways to deploy an Artie:
- Cloud
- You own, operate, and maintain Artie's hardware
- Artie compute required for administration is provided by the cloud
- Any additional compute required for experiments or workload is provided by the cloud
- See here for how to get started with this route
- Fog
- You own, operate, and maintain Artie's hardware
- Artie compute required for administration is owned, operated, and maintained by you in the same network as Artie
- Any additional compute required for experiments or workload is provided by the cloud
- See here for how to get started with this route
- Edge
- You own, operate, and maintain Artie's hardware
- Artie compute required for administration is owned, operated, and maintained by you in the same network as Artie
- Any additional compute required for experiments or workload is provided by you locally in the same network as Artie
- See here for how to get started with this route
Note that since Artie is so early in development, none of these are viable options yet! Additionally, Edge is the highest priority for development, while the other options are ones I hope to get to, but may not.
Here are a few links to architectural discussions:
- Architecture Overview for the whole System
- Architecture Overview for the Robot Itself
- Security Overview
- CAN Protocols - We overlay several protocols on top of CAN. This document describes them in detail.
- MsgPack Schema - We use MsgPack for some of the serialization/deserialization.
TODO: We should probably have a 1:1 mapping of Artie bots to Elasticsearch nodes.
Developmental robotics is the study of human development by means of simulating it. It is an important field of study for at least the following two reasons:
- Developmental robotics informs the study of human development. What better way to test a theory of how a human develops than to try to build a human?
- Developmental robitics informs the study of artificial intelligence. Although not typically the main focus of developmental robotics, AI can benefit from any advances in our understanding of human intelligence.
Here's a great excerpt from Wikipedia:
As in human children, learning is expected to be cumulative and of progressively increasing complexity, and to result from self-exploration of the world in combination with social interaction. The typical methodological approach consists in starting from theories of human and animal development elaborated in fields such as developmental psychology, neuroscience, developmental and evolutionary biology, and linguistics, then to formalize and implement them in robots
Developmental robotics is such a cool field. I wish way more people were interested in it.
A central tenet of developmental robotics is that embodiment (agency) is necessary for intelligence. Human intelligence stems from a need to manipulate the world and ourselves in response to the world. It does not exist in a vacuum. Hence, it makes sense to embody an intelligent agent in a physical (or at least virtual) world.
If developmental robotics is the study of human development by means of simulating it, why build an actual robot? Why not just simulate one in a physics simulator?
The answer is simple: humans aren't simulated - they exist in a messy, physical world. Any theory of how a human develops must ultimately be put to the test in the same world with the same parameters.
Nonetheless, there is a place for virtual simulation in developmental robotics, and Artie will hopefully incorporate a simulator, since working with an actual robot is way less convenient than working in a video game.
One last thing about robots: embodiment is a two-way street. Though you can get by to an extent with datasets that are impersonal and aggregated (as in the typical supervised learning paradigm of machine learning), humans do not learn this way. Humans learn from interacting with their environment and by their environment interacting with them. Parents speak directly to their children using infant-directed speech. Teenagers navigate social environments that are unique to their particular group of friends. Young adults take elective classes at university that are interesting to them. An intelligent agent has an active, often causative relationship with the data from which it learns. Hence, you need to place the subject into an environment to truly study the development of intelligence.
Also, robots are awesome!
A few reasons:
- They're so expensive. Holy crap are they expensive:
- Affordable ones cannot feasibly be used to study human development:
Artie is built from the ground up explicitly for the purpose of developmental robotics. He's also open source, and as cheap as is feasible (though it turns out, that's still pretty expensive).