Robot Simulation
Robot Simulation environments are the only way to train and test robot control systems to the limit.
We create and build on all major robot simulation environments.
We work with simulators to:
- Develop and improve sensor simulation models
- Prototype robot controllers
- Apply Reinforcement Learning to learn behaviors
- Visualize, re-process and analyse real-world data captures
- Use physics engines to generate dynamic environments
- Model dynamical systems
- Simulate Multibody Kinematics and Dynamics
Accelerate Task programming with AI
As a Preferred Solution partner of NVIDIA, we have the access to and experience with the NVIDIA AI stack for robot programming. Robot, environment and goals are defined in the Isaac Lab simulation framework, leading to a Convolutional Neural Network (CNN) capable of executing the task, even in the presence of disturbances.
This is achieved through domain randomization (ie Generative AI for Synthetic Data Generation) and millions of training cycles executed on affordable GPUs.