AI Development and Modeling

in: automation, robots

We use artificial intelligence (AI) at Robots Everywhere, and with the uncertainties around the safety and ethics of AI programs, we wanted to share some clarity with everyone.

Financial Automation

Algorithmic trading has been a growing part of Robots Everywhere’s financial backend, running a unique algorithm that we have developed entirely in-house. The algorithm is compatible with a wide variety of large language models and processing intents. Through a combination of backtesting on similar markets and real-time trading, we have developed an algorithm that is consistently profitable on both US and Canadian markets. While licensing it to other parties is still under legal review, we are looking to open the door to private partners in the algorithmic trading and fintech spaces.

So far, the most measurable advantages of the system we have developed compared to alternatives in the space are energy efficiency and security. By keeping its data processing entirely local, it does not rely on huge amounts of cloud computing power to function, nor does it expose any of its communications. The former helps ensure reliable profit margins above energy and equipment costs, and the latter reduces the risk of a disruptive cyberattack.

Edge Computer Vision Models for Drones

Edge AI is defined as artificial intelligence models running on, or locally networked to, an operating device. In the drone world, this means developing AI models that are optimized for size and performance. A small drone or a ground control station will not have a large, triple-digit-watt GPU. They rely on battery power, and even their high consumption companion computers have at most a few gigabytes of memory. Can computer vision tasks be performed on these machines in order to guide precision landings, identify objects, or perform autonomous search and rescue? Yes, they can.

More than that, we have even made breakthroughs in running AI models directly on microcontrollers such as the ESP32 and Parallax P2, moving AI off the edge, and onto the device. We plan on showcasing more of these developments in 2025, in a variety of spaces.

For more edge AI solutions, contact us.

AI Generated Media Content Policy

Robots Everywhere has a commitment to honesty and quality. Feedback loops, where AI models feed new data into other AI models to generate media content, have become a singificant problem in news and social media today. Robots Everywhere is committed to standing against that

While we do make use of AI in content generation, human writers and editors are still the first and last step in all of our media process. We have no plans to change that in the near future, not until we have developed an LLM with a vetted training data set that can demonstrate a sufficient degree reliability and safety. Should we develop such a project in the future, it will be used to assist human writers, not replace them, and will be trained with ethical use cases in mind.

Commitment to Sustainability

With vast AI data centers becoming more common, with power generation requirements that rival cities rather than businesses, we know we have to make a statement on this subject. Robots Everywhere is committed to building the same way we always have: lean and efficient. Our AI solutions will continue to be focused on local execution and minimizing the need for computing resources. Our commitment to the long-term future of humanity means we will not support low efficiency AI that uses scale as the only answer to every problem. We respect the principle of diminishing returns, and continue to expect the same of others.