UNER.
VHIL / HARDWARE-IN-THE-LOOP WITH VIRTUAL SENSORS
Connect Radar, LiDAR, Camera and Powertrain target boards. UNER supplies virtual sensor data and simulated vehicle-state inputs. AI, sensor processing and powertrain control run on the attached hardware; results and control outputs return to the simulation.
Physical AI / AI Development / Safety Embedded
Interactive concept illustration · not a live simulator feed
VHIL / A CLOSER LOOK
Explore the AMR vehicle simulation, the Physical AI workflow and the radar hardware demonstration. Our VHIL setup spans Radar, LiDAR, Camera and Powertrain target boards.
Follow a continuous simulated drive through vehicle features, ending with a combined radar and LiDAR view of the surrounding scene.
Download the videoA look at the current simulation build: road scenes, virtual radar and vehicle telemetry across changing driving conditions.
Download the previewA visual introduction to the journey from simulated environments to embedded intelligence.
Explore radar coverage in the simulated environment. UNER supplies virtual radar inputs to real target hardware and brings processing results back into the simulation for comparison and analysis.
Explore more radar viewsVIRTUAL RADAR / SCENE TO SENSOR
Explore short recordings of simulated driving, radar detections, sensor coverage and combined radar and LiDAR views. Choose a view, then expand it to inspect the details.
DRIVING SCENARIOS
Follow the simulated road scene alongside the radar display.
CHOOSE A VIEW 10 recordings
Five-second simulation recordings ? select a thumbnail and press play ? use fullscreen for detailed inspection.
OUR ENGINEERING CAPABILITIES
Work with us on a focused engineering challenge or connect the full development chain: model the system, develop its intelligence and integrate the software that runs it.
Use simulation as a virtual sensor source for attached hardware. Evaluate the target’s AI and embedded software against repeatable physical scenarios.
Turn sensor data into models, then evaluate their quality and prepare them for the constraints of the intended target.
Engineer embedded software around defined operating limits, supervised control, fault handling and verifiable behaviour on the target hardware.
The physical task defines the requirements. AI and control algorithms implement the behaviour. Safety Embedded connects execution, supervision and verification on the target.
01 / PHYSICAL AI — THE SIMULATION TOOLKIT
Autonomous vehicle development brings our capabilities together. Explore vehicle physics, LiDAR, camera and radar, then connect them around the same environment and development task.
VHIL / THE CONNECTED SYSTEM
Virtual sensor data feeds Radar, LiDAR and Camera target boards. The vehicle model provides state feedback to the Powertrain target board. Processing results and control outputs return through the configured interfaces for evaluation and vehicle response.
DEVELOPMENT FOCUSSensor processing and powertrain control on real hardware
Simulink-generated vehicle dynamics connect powertrain, steering, braking, tyres and suspension to the simulated road.
Commands → physical responseExplore scan patterns, point returns and scene geometry with configurable sensor placement and ray-based sensor models.
Scene geometry → point cloudsCapture the Unreal scene from the vehicle’s perspective. Prepare images for perception and synthetic data workflows.
Rendered scene → image framesConnect modelled reflectors to FMCW signal processing, range–Doppler representations and detection pipelines.
Modelled returns → detectionsVHIL / FROM SIMULATION TO REAL HARDWARE
Radar, LiDAR and Camera boards receive virtual sensor data. The Powertrain board exchanges state feedback and control outputs with the simulated vehicle. Each hardware path connects execution to its physical context.
Road geometry, traffic participants, lighting and environmental conditions.
Send camera images, LiDAR points and radar data to sensor targets, and vehicle-state feedback to the Powertrain target.
Run sensor processing, AI inference and control on Radar, LiDAR, Camera and Powertrain target boards.
Compare target outputs with the scenario. Where connected, apply control commands to the vehicle model.
Hardware results → evaluation Connected control → vehicle response → next observation
From a component question to a system question.
How does a sensor configuration affect the perception input? How does a delayed decision change the vehicle’s response?
02 / AI DEVELOPMENT
Develop the model and its execution path together. Our radar CNN workflow connects data preparation, training, optimization and native inference integration.
See the radar development demoOrganize sensor datasets, inspect labels and scenario coverage, and separate training, validation and test data.
Inspect detection quality and failure cases. Explore compact architectures, pruning and INT8 quantization against the task’s requirements.
Export model artifacts, connect native inference and compare outputs with the reference implementation. Measure timing and resources on the intended target.
03 / SAFETY EMBEDDED
Build safety-oriented embedded functions with explicit operating limits, input validation, command supervision and defined fault responses. Connect requirements to implementation and verification on the intended target.
Explore Mobile Robot ArchitectureC/C++ components, startup integration and RTOS task organization, with explicit timing, memory and interface constraints.
Validate sensor inputs and command limits, manage operating-mode transitions, and define responses to stale data, communication loss and detected faults.
Connect requirements, software components and test scenarios. Check boundary conditions, fault responses, execution timing and diagnostics on the intended target.
Engagements can focus on one discipline or combine all three. Deliverables, target platforms and acceptance criteria are defined for each project.
MODULAR BY DESIGN
The SGPN SDK — Safety Guardian Perception & Navigation — provides the C/C++ foundation behind this direction. Reusable components connect sensor processing, autonomy interfaces and the physical vehicle model.
Unreal supplies the scene. Simulink supplies the vehicle equations. VHIL adapters connect virtual sensors to Radar, LiDAR and Camera boards, and the vehicle model to the Powertrain board. Results and control outputs return to the development environment.
Explore Mobile Robot ArchitectureAn evolving engineering platform. Vehicle dynamics and sensor implementations are available in the development workspace. Integration maturity varies by module; localization, fusion, autonomy and hardware targets require project-specific implementation and validation.
YOUR NEXT ENGINEERING PROJECT
Physical AI, AI development or safety embedded engineering. Tell us about your application, target hardware and the next development decision you need to make.
