One Robot. Hardware, Simulation, and a Digital Twin.
Efferax builds the software that makes robots autonomous — navigation, fleet management, sensor fusion, and agentic behavior planning — for equipment inspection, autonomous driving, and logistics. Every platform we ship exists physically, inside Unreal Engine 5, and as a live digital twin. Toggle the viewer to see all three.
- Production sensor stack
- Edge GPU autonomy node
- IP67 field-rated chassis
Every robot ships in three realities
Buy the hardware, validate in Unreal Engine, operate through the digital twin — all running the same autonomy software.
Hardware
Field-rated robots with production sensor stacks and edge GPU autonomy nodes running the full ROS 2 stack.
- IP67 chassis, -20°C to 55°C
- Edge GPU 100–275 TOPS
- Hot-swap batteries, pilot kits
Unreal Engine 5 Simulation
Photoreal digital environments with ray-traced synthetic sensors so you can test missions before a robot ever leaves the lab.
- Lumen & Nanite scenes
- Synthetic LiDAR, RGB-D, thermal, radar
- Scenario builder & replay
Live Digital Twin
A continuously synchronized twin of every deployed robot — joint states, pose, sensor health, and mission events at up to 1 kHz.
- Real-time telemetry mirror
- Predictive maintenance
- Frame-by-frame mission replay
The closed Sim-to-Real-to-Twin loop
Scenarios are born in Unreal Engine 5, behaviors are verified before deployment, and every field mission flows back into simulation through the digital twin.
Unreal Engine 5 Scenario
Photoreal substations, warehouses and roads built with Lumen & Nanite. Weather, lighting, and obstacle density are parameterized for domain randomization.
UE5 • Chaos PhysicsSynthetic Sensor Data
Ray-traced LiDAR point clouds, stereo RGB-D, radiometric thermal, and radar returns generated with ground-truth labels at scale.
LiDAR • RGB-D • ThermalTrain & Verify Behaviors
Perception models and agentic behavior trees are trained, fuzzed against thousands of edge cases, and formally checked by the safety firewall.
BT.CPP • CBF VerificationDeploy to Hardware
The identical ROS 2 autonomy stack runs on the edge GPU of the physical robot. Same code, same interfaces, zero sim-only shortcuts.
ROS 2 • Edge GPULive Digital Twin
Every deployed robot streams joint states, pose, sensor health, and mission events into a twin you can monitor, replay, and query.
SSE • 1 kHz TelemetryClose the Loop
Field anomalies become new simulation scenarios automatically, so the fleet keeps learning from every mission.
Continuous LearningA heterogeneous fleet, one shared world model
Aero-V flies overwatch and maps the site, Quad-I patrols and scans assets, AMRs move material through reserved corridors, Drive-S runs the perimeter, and Arm-M handles contact inspection — all coordinated by the fleet engine and mirrored in a single digital twin. Watch the drone hand a thermal anomaly to the quadruped and the AMRs negotiate the shared zone without a deadlock.
Four pillars of the Efferax Autonomy Stack
Autonomous Robot Navigation
3D SLAM • Motion Planning • Autonomous Driving
Full-stack spatial autonomy for GPS-denied facilities and outdoor vehicles: tightly-coupled LiDAR-Inertial-Visual SLAM paired with real-time optimal control.
- 3D LiDAR-Inertial Odometry (FAST-LIO2 / LIO-SAM) with loop closure
- Theta* / Hybrid A* kinematically feasible global search
- MPPI & Model Predictive Control at 200+ Hz
- Spatio-temporal voxel layers for dynamic obstacles
Robot Fleet Management
Multi-Agent Coordination • Traffic Deconfliction • Telemetry
Cloud-to-edge fleet command that orchestrates heterogeneous robots across shared workspaces without deadlocks or single points of failure.
- Auction-based task allocation & battery-aware dispatching
- Time-expanded graph routing for deadlock-free corridors
- VDA5050 & Open-RMF interoperability
- Low-latency bi-directional telemetry & remote intervention
Sensor Data Processing
Multi-Modal Fusion • 3D Point Clouds • Equipment Diagnostics
High-throughput perception pipelines turning LiDAR, stereo RGB-D, acoustic and radiometric thermal streams into actionable 3D semantic twins.
- Real-time point cloud deskewing, segmentation & voxelization
- Hardware-synced stereo + thermal projection
- Gauge OCR, hotspot & corrosion defect detection
- EKF & factor-graph multi-sensor state estimation
Agentic AI Behavior Planning
Cognitive Reasoning • Dynamic Behavior Trees • Safety Guards
Multi-agent AI interprets unexpected mission events and synthesizes verified Behavior Trees — bridging rigid state machines and open-world uncertainty.
- LLM/VLM semantic reasoning → BT.CPP execution
- Dynamic subtree synthesis when nominal plans fail
- Deterministic Control Barrier Function safety firewall
- Episodic memory for fleet-wide continuous learning
The Efferax lineup
Efferax Quad-I
Legged Inspection Platform for Substations, Refineries & Turbine Halls
Efferax AMR-L
Fleet-Native Autonomous Mobile Robot for Logistics & Intralogistics
Efferax Aero-V
Autonomous Inspection Drone for Aerial Survey, Thermal Mapping & Fleet Overwatch
Efferax Arm-M
Mobile Manipulation Platform for Contact Inspection & Intervention
Efferax Drive-S
Autonomous Shuttle & Yard Vehicle Stack for All-Weather Mobility
How Agentic AI Makes Robot Behavior Planning Smarter
Static state machines break on the first unforeseen edge case. Pick a field anomaly and watch the live Python robotics-engine reason over sensor context and synthesize a verified BehaviorTree.CPP / ROS 2 subtree in real time — the same plan that would run on hardware, in Unreal simulation, and against the digital twin.
Select a scenario or describe a custom mission event, then click "Synthesize Behavior Plan" to stream a live Behavior Tree and subsystem verification from the Python Robotics Engine.
<BehaviorTree ID="AdaptiveInspectionSubtree">
<Sequence name="ThermalAnomalyVerification">
<Action ID="HaltNominalPatrol" deceleration="1.2m/s2" />
<Action ID="AlignPanTiltGimbal" target="flange_hotspot_roi" />
<Parallel success_threshold="2">
<Action ID="CaptureRadiometricThermalBurst" frames="16" />
<Action ID="PointCloudHighResScan" voxel_size="0.002m" />
</Parallel>
<Action ID="PublishInspectionAlert" severity="CRITICAL" />
<Action ID="ResumeFleetRoute" />
</Sequence>
</BehaviorTree>Engineered for high-stakes physical environments
Autonomous Equipment Inspection
Quadrupeds and tracked AMRs patrolling substations, refineries and turbine halls — thermal hotspot detection, gauge OCR, acoustic leak localization, and 3D change detection between rounds.
Autonomous Driving & Off-Road
All-weather perception (3D LiDAR, 4D radar, stereo) and MPPI/MPC trajectory planning for shuttles, yard tractors, and unstructured field vehicles under degraded visibility.
Fleet Logistics & Intralogistics
VDA5050-native AMR swarms with deadlock-free traffic control, battery-aware scheduling, and a shared semantic map — monitored through one fleet-wide digital twin.
Start in simulation. Deploy on hardware. Operate through the twin.
Join the Efferax pilot program to get early access to the Unreal Engine simulation sandbox, digital twin console, and hardware evaluation units.