What is Covariant?

Covariant is an AI-based software company that specializes in products used to teach robots new skills. The company, which is based in Berkeley, California, was founded in 2017 by former OpenAI employees Pieter Abbeel and also Peter Chen, Tianhao Zhang, and Rocky Duan.

A short history of Covariant

Abbeel and his team started Covariant because they wanted to use their AI research to solve problems in real-world scenarios characterized by constant change and infinite variability. 

The first problem Covariant tackled was to help robots learn how to pick up objects of various sizes, shapes, colors, and textures. While humans take the ability to grasp different objects for granted, teaching a robot to perform the task was more complex with many failed attempts over the last decade or so.

As a consequence, a robot that could replicate a human worker was extremely desirable among warehouse and logistics companies looking to automate. Based on this need, Covariant developed an AI-based system that trained network robots to pick up various items and improve their picking ability over time. 

Future potential

This was the first general skill the robots learned, but Chen later explained in a TechCrunch interview that their ability to understand 3D space and objects would be useful in a plethora of other applications.

Take a robot that picks watermelon, for example. Once that robot has learned to identify and grasp a watermelon at a packing facility, it can also pick watermelons off the vine at the farm on which they are grown. 

Chen also explained a tech trend that was currently occurring at the intersection of AI, software, and mechatronics. While mechatronics had traditionally been distinct from the other two, he noted that modern businesses equated the intersection with reliability because it enabled machines to do “the same thing over and over again” without mistake. 

Covariant Brain

Covariant Brain is the name of the company’s universal AI platform for warehouse picking operations. It reached human-level autonomy in early 2018 and was purchased by leading warehouse integrators such as ABB, Knapp, and Bastian twelve months later. 

It was also in 2019 that Covariant Brain won an ABB order-picking competition as the only entrant in a field of 20 robots that solved all 26 complex challenges. Some of the characteristics of Covariant Brain include:

  • Fleet learning – Covariant Brain continuously improves because it learns from any connected robot.
  • Versatility – the system can pick almost any item or stock-keeping unit (SKU) – irrespective of its packaging, material, size, or shape.
  • Ready-to-go – Covariant Brain is also pre-trained on the data from millions of robot picks around the world.

Covariant team

To conclude, let’s take a slightly more detailed look at the company’s four co-founders.

Pieter Abbeel

Pieter Abbeel is a Belgian-born computer scientist and electric engineer who is best known for his innovative research in machine learning and robotics. 

Abbeel is the current director of the Berkeley Robot Learning Lab and co-director of Berkeley AI Research (BAIR). He was also one of the early employees of OpenAI.

Peter Chen

Peter Chen is an American computer scientist and former research scientist at OpenAI who is currently Covariant’s CEO. Whilst at Berkeley under Abbeel, Chen co-created a deep unsupervised learning course.

Tianhao Zhang

Tianhao Zhang also worked under Abbeel at Berkeley while he completed his Ph.D. in computer science between 2016 and 2021. 

Before co-founding Covariant, Zhang worked as a research intern at Berkeley and Microsoft and has authored papers on imitation learning and deep control policies for autonomous drones.

Rocky Duan

Yan (Rocky) Duan was the third of Abbeel’s Ph.D. students at Berkeley. His thesis focused on meta-learning for control, which involves “policy learning algorithms that can themselves generate algorithms that are highly customized towards a certain domain of tasks.

Later, Duan became one of OpenAI’s early hires at just 21 years of age.  

Key takeaways:

  • Covariant is an AI-based software company that specializes in products used to teach robots new skills. The company was founded in 2017 by former OpenAI employee Pieter Abbeel and also Peter Chen, Tianhao Zhang, and Rocky Duan.
  • Covariant Brain is the name of the company’s universal AI platform for warehouse picking operations. It reached human-level autonomy in early 2018 and was purchased by leading warehouse integrators such as ABB, Knapp, and Bastian twelve months later.
  • Covariant’s product was developed in response to a trend Chen believes is occurring at the intersection of AI, software, and mechatronics. Businesses looking to automate warehouse operations desire multi-functional machines that can pick reliably and consistently.

Read Next: History of OpenAI, AI Business Models, AI Economy.

Connected Business Model Analyses

AI Paradigm

current-AI-paradigm

Pre-Training

pre-training

Large Language Models

large-language-models-llms
Large language models (LLMs) are AI tools that can read, summarize, and translate text. This enables them to predict words and craft sentences that reflect how humans write and speak.

Generative Models

generative-models

Prompt Engineering

prompt-engineering
Prompt engineering is a natural language processing (NLP) concept that involves discovering inputs that yield desirable or useful results. Like most processes, the quality of the inputs determines the quality of the outputs in prompt engineering. Designing effective prompts increases the likelihood that the model will return a response that is both favorable and contextual. Developed by OpenAI, the CLIP (Contrastive Language-Image Pre-training) model is an example of a model that utilizes prompts to classify images and captions from over 400 million image-caption pairs.

OpenAI Organizational Structure

openai-organizational-structure
OpenAI is an artificial intelligence research laboratory that transitioned into a for-profit organization in 2019. The corporate structure is organized around two entities: OpenAI, Inc., which is a single-member Delaware LLC controlled by OpenAI non-profit, And OpenAI LP, which is a capped, for-profit organization. The OpenAI LP is governed by the board of OpenAI, Inc (the foundation), which acts as a General Partner. At the same time, Limited Partners comprise employees of the LP, some of the board members, and other investors like Reid Hoffman’s charitable foundation, Khosla Ventures, and Microsoft, the leading investor in the LP.

OpenAI Business Model

how-does-openai-make-money
OpenAI has built the foundational layer of the AI industry. With large generative models like GPT-3 and DALL-E, OpenAI offers API access to businesses that want to develop applications on top of its foundational models while being able to plug these models into their products and customize these models with proprietary data and additional AI features. On the other hand, OpenAI also released ChatGPT, developing around a freemium model. Microsoft also commercializes opener products through its commercial partnership.

OpenAI/Microsoft

openai-microsoft
OpenAI and Microsoft partnered up from a commercial standpoint. The history of the partnership started in 2016 and consolidated in 2019, with Microsoft investing a billion dollars into the partnership. It’s now taking a leap forward, with Microsoft in talks to put $10 billion into this partnership. Microsoft, through OpenAI, is developing its Azure AI Supercomputer while enhancing its Azure Enterprise Platform and integrating OpenAI’s models into its business and consumer products (GitHub, Office, Bing).

Stability AI Business Model

how-does-stability-ai-make-money
Stability AI is the entity behind Stable Diffusion. Stability makes money from our AI products and from providing AI consulting services to businesses. Stability AI monetizes Stable Diffusion via DreamStudio’s APIs. While it also releases it open-source for anyone to download and use. Stability AI also makes money via enterprise services, where its core development team offers the chance to enterprise customers to service, scale, and customize Stable Diffusion or other large generative models to their needs.

Stability AI Ecosystem

stability-ai-ecosystem

About The Author

Scroll to Top
FourWeekMBA