Inverted AI, a Vancouver-based company that uses generative artificial intelligence for autonomous vehicle development, announced Thursday it has raised $4 million in early-stage funding to realistically simulate human drivers on roadways.

The seed funding round was led by Yaletown Partners, with participation from investors including Blue Titan Ventures, Dasein Capital, Inovia Capital, Defined and WUTIF.

Inverted AI creates technology that simulates roadways, vehicles and drivers using what the company calls “nonplayer characters” or NPCs controlled by generative AI agents that mirror human behavior. The infusion of capital will go toward the flagship simulation system of the company called Imagining the Road Ahead, or ITRA, a foundation model that allows the company to provide behavioral models for the safe implementation of autonomous solutions on roads.

“Inverted AI’s predictive human behavioral models are designed to bring the most realistic behavior to simulation based on massive quantities of video data, that result in simulations with reactive, diverse and realistic NPCs across vehicle classes and pedestrians,” said Chief Executive Frank Wood.

Using ITRA, autonomous vehicle developers can deploy maps of numerous different scenarios and include NPCs reactive humanlike drivers that will react and act like human actors in similar situations. These maps can be real-world locations or fictitious places, such as locales or edge cases designed to put robot vehicles through their paces.

Each agent can only “see,” as in its literal view of the world, a local rendering of the environment around itself. That means its awareness is tied to what an actual human in these situations would experience and react to, including blind spots caused by bushes or sharp angles. This enhances their behavior and allows the Inverted AI team the ability to change how the maps are generated for future feature updates.

The company currently offers an application programming interface that allows autonomous vehicle developers to access scenarios through a set of different products. The company’s Drive service provides AI agent-controlled drivers with simulation states and behavior controls. It currently supports only vehicles, but pedestrians will be added in the future.

Initialize allows developers to quickly place AI-controlled agents in the simulated world as if they were naturally placed and already driving. Scenario allows for the generation of entire scenarios including cars, buses, pedestrians and lights. Finally, Blame will determine which agent caused a collision.

“The opportunity of generative AI to advance adaptive systems and accelerate the convergence of simulation and real-world applications, in our view, is one of the great thematic changes underway,” Eric Bukovinsky, a partner at Yaletown Partners. “Inverted AI is an extraordinary team that has created the world’s leading foundational models in generative AI for human behavior.”

The company said that it would use the new funding to expand its research and product offerings, especially related to its ITRA technology to accelerate the commercial deployment of autonomous solutions on roadways.

Inverted AI also recently launched a new site with an interactive demo of the API that allows researchers to sign up for a trial key so that they can test out their own robots and autonomous vehicles using the system.

Image: Inverted AI