Onyx Robotics :: AI-powered Tooling for Robotics Controls & Physics Simulation

Onyx Robotics is a Boston-based early-stage startup specializing in AI-powered tooling for robotics controls and physics simulation. The company develops a cloud platform that enables engineering teams to create highly accurate digital twins and simulation models from real-world hardware sensor data, reducing or replacing expensive and time-consuming physical testing.

Onyx focuses on the “tooling layer” for AI controls and simulation. Its core offering helps robotics and hardware teams model complex real-world dynamics (non-linear behaviors, environmental interactions, sensor noise, etc.) using time-series data from existing hardware. This allows for:

• Accurate Physics Simulation: Trusted digital twins that closely match real hardware performance, enabling extensive testing in simulation.
• Zero-Shot Transfer: Deploying advanced control algorithms directly to physical robots with minimal real-world tuning.
• Faster Engineering Cycles: Significant reductions in iteration time and costs associated with physical prototyping and testing.

Key technical features include:

• Training AI models on proprietary or customer-provided sensor data.
• Exporting models to industry-standard libraries in open-source code for local, real-time execution.
• One unified workflow supporting multiple hardware systems.
• High-performance implementations optimized for controls research and production use.

The company positions itself as infrastructure for next-generation robotics development, helping teams push hardware performance limits through better simulation fidelity and control strategies. It is a resident startup at the MassRobotics incubator in Boston and is backed as a venture capital-supported entity. Leadership includes Founder & CEO Ted Lutkus and COO Kyung Keun Park.

While not manufacturing physical robots, Onyx Robotics builds essential software tools that accelerate the development and deployment of AI-controlled robotic systems across various applications. Their solution targets controls and simulation teams working on complex autonomous systems where real-world testing is a major bottleneck. The platform emphasizes practicality — models run locally with high performance and integrate into existing engineering stacks.

As a small team (reported as 1–10 employees), Onyx is in the early growth phase, actively hiring for roles like Robotics Research Lead. They emphasize open collaboration and are building tools inspired by leading AI controls research to bridge the sim-to-real gap more effectively.