Antioch startup develops simulation tools for robotics

Antioch, a startup focused on creating simulation tools for robot developers, aims to close what is known as the sim-to-real gap. This challenge involves creating virtual environments that are realistic enough for robots trained within them to operate successfully in the physical world. Antioch's CEO and co-founder, Harry Mellsop, announced that the company has raised $8.5 million in a seed funding round, valuing it at $60 million.

Mellsop emphasized that improving simulation is crucial for the advancement of physical AI, which is a key focus for many major autonomy companies. For instance, Waymo uses Google DeepMind's world model to test its vehicles. However, creating simulations for robots requires different skills compared to cars, and Antioch aims to provide a platform that assists new companies lacking the capital to build their own testing arenas.

He also pointed out that the vast majority of the industry currently does not utilize simulation, and this needs to change. Antioch compares its products to Cursor, a popular AI-powered software development tool, allowing robot builders to create multiple digital instances of their hardware and connect them to simulated sensors.

The challenge lies in ensuring that the physics in the simulation matches reality to prevent issues when the model controls a real machine. Antioch starts with models built by companies like Nvidia and develops domain-specific libraries to simplify their use. Collaborating with multiple clients enables the company to accumulate context for refining its simulations.

Currently, Antioch's focus is on sensor and perception systems, which are critical for automated cars, agricultural and construction machinery, and drones. While the startup targets smaller companies, some of its early clients include large multinationals heavily investing in robotics.