Motphys Debuts MotrixSim 1.0 Simulation Engine and Forge Scene Tool at CoRL 2026

Motphys unveiled MotrixSim 1.0, its fully in-house embodied intelligence simulation engine, at CoRL 2026, claiming a 12.8× speedup over an Isaac Sim-based solution on a tracking task and a 100% pass rate in a grasping test. PR Newswire published the announcement on November 14, 2026.

Inside the MotrixSim 1.0 Benchmarks

Motphys built the engine's core from the ground up in Rust rather than on existing open-source engines, which the company said gives MotrixSim 1.0 memory safety and computation power. In a disturbance-rejection grasping test where a Franka arm grasps objects of different shapes, MotrixSim 1.0 was the only engine to achieve a 100% pass rate at both 2 ms and 10 ms simulation step sizes, according to the company. In a critical friction point test, the engine showed hold-and-slip behavior near the critical friction consistent with Coulomb's law, with a 0mm slip distance.

Through a pipeline supporting CPU/GPU parallel computation, MotrixSim can reach a physics simulation throughput of millions of physics steps per second across 4,096 parallel environments. In a controlled comparison on the G1 whole-body tracking task, under the same task environment configuration and the same reinforcement learning algorithm, MotrixSim reached the current tracking score in about 10.2 minutes of end-to-end time, while the Isaac Sim-based solution took 130.8 minutes. For the Strandbeest complex closed-loop mechanism with 92 linkages, MotrixSim solves closed-loop constraints together with multibody dynamics in generalized coordinates, keeping the mechanism's assembly relationships during motion at a 5 ms time step where traditional engines failed to maintain stability.

MotrixSim 1.0 also reports LiDAR simulation throughput of 124 million rays per second and stable visual synthesis throughput when resolution rises 16 times. The company cites high-speed solving in a single line of code with a 98.42% success rate and a "zero-conversion + AI auto-modeling" workflow. Motphys also launched Forge, described as the industry's first Blender-based extension tool, which lets AI agents generate physics scenes automatically.

"Simulation once served to validate how machines perform. Now, it is where they learn how to perform," said Zhang Heng, co-founder and CTO of Motphys. Zhang said the core proposition of Physical AI is to turn expensive and slow real-world physical experience into digital learning resources that can be generated at scale. Motphys's research has been accepted to conferences including RSS and CoRL, and the company has built a simulation and real-robot training center for embodied AI in China. The company names Google's MuJoCo and NVIDIA's Isaac Sim among the leading physics engines worldwide.

PaXini and Pudu Launch Robotics Tools

Other robotics firms have also introduced products in recent weeks. PaXini AI showcased tactile sensing and its TORA-series humanoid robots at RobotWorld 2026, presenting its PIE full-stack architecture for Physical AI deployment, The future of trading reported on November 7, 2026. PUDU Robotics launched a self-sharpening lawnmower robot in its TOMO S series at its first European event and announced a unified software architecture, Forbes reported on November 9, 2026. Serve Robotics launched Physical AI Services to sell connectivity and data tools to other robot builders, TechCrunch reported on November 5, 2026.