PROJECT OVERVIEW
Looking Glass XR partnered with FS Studio to build the simulation foundation for the TRON 1 bipedal patrol robot. The objective was to create a full digital twin inside NVIDIA Omniverse that could support realistic locomotion, sensor simulation, patrol behaviors, and future computer vision training for analog gauge detection.
This work represents Phase 1 of a broader roadmap toward a deployable robot as a service platform. The focus was not visualization. The focus was functional simulation that could support real robotics development and future autonomy.
THE CHALLENGE
The initial TRON 1 robot model was not operational in a meaningful way.
- The robot could not move or even self-balance
- There was no validated control stack
- Navigation inside a 3D environment was not possible
Without a reliable simulation stack, progress on hardware would require repeated trial and error on the physical robot. That approach would be slow, risky, and expensive. The client needed a simulation environment where behavior could be rebuilt, tested, and validated before moving forward.
THE APPROACH
FS Studio designed and implemented a simulation first workflow using NVIDIA Omniverse, Isaac Sim, and Isaac Lab. The goal was to rebuild the robot’s behavior from the ground up and create a stable foundation that could scale into future phases.
Key steps included:
- Building a fully containerized Isaac Sim environment to support reproducible builds and external configuration
- Importing the TRON 1 robot into a USD based Omniverse pipeline
- Integrating ROS 2 for all command inputs and telemetry outputs
- Simulating RealSense style sensors and exposing their data through ROS 2 topics
- Creating a configuration system for Points of Interest and analog gauge spawning
- Retraining the robot model using Isaac Lab and LymX resources to improve balance and movement
- Validating controlled navigation inside a realistic electrical room environment
This approach ensured that physics, sensors, control, and training all operated within a single consistent simulation environment.
THE SOLUTION
FS Studio delivered a trained and operational TRON 1 digital twin inside NVIDIA Omniverse.
Isaac Lab was used to retrain the robot for stability and movement. This allowed the team to iterate on balance, responsiveness, and control loops without putting physical hardware at risk.
Why this stack mattered:
- Isaac Sim provided physically consistent articulation and contact dynamics required for bipedal stability
- Isaac Lab enabled structured training loops and repeatable experiments
- The Omniverse environment ensured all components shared the same source of truth
This combination turned the robot from a static model into a controllable and testable system.
RESULTS
The updated TRON 1 simulation now supports:
- Stable self balancing
- Keyboard driven manual control
- Forward and backward motion
- Left and right rotation
- Controlled navigation inside an electrical room scene
- Point of Interest spawning and directed movement toward defined coordinates
The robot behaves consistently and predictably inside simulation, enabling meaningful development and testing.
WHAT THIS ENABLES NEXT
With the Phase 1 foundation complete, the client can now expand the platform without reworking the core system.
Future capabilities include:
- Autonomous navigation between predefined Points of Interest
- More complex task sequencing trained in Isaac Lab
- Higher level patrol logic layered on top of the existing movement stack
- Integration with Nav2 for navigation planning
- Computer vision pipelines for analog gauge detection
- Sensor driven autonomy and perception based decision making
Each of these builds directly on the existing simulation environment.
TOOLS AND TECHNOLOGIES
- NVIDIA Omniverse for USD based robot and environment modeling
- Isaac Sim for physics, articulation, and sensor simulation
- Isaac Lab for training and stability improvement
- ROS 2 integration for control and telemetry
- Containerized Omniverse workflows for reproducibility
CLIENT READY SUMMARY
FS Studio delivered a functional ROS 2 powered digital twin of the TRON 1 patrol robot inside NVIDIA Omniverse. The robot now balances, moves, and navigates virtual environments reliably. The system supports sensor simulation, Points of Interest, and training workflows, and it establishes a scalable technical foundation for future autonomy and computer vision driven patrol capabilities.