Embedded Inertial Sensing ForPhysical AINative to your robot
Multi-Redundant MEMS. Onboard olixAI™ fusion. Native ROS 2. Relative or magnetometer-aided orientation, angular velocity and acceleration—connected and powered through one Ethernet cable.
Connect motion sensing directly to your robotics stack. X2 Ultra publishes standard ROS 2 messages from the device, with timestamps, frame identifiers and covariance fields.
Build around familiar topics for orientation, inertial measurements, magnetic field, temperature and diagnostics. Use TCP or UDP JSON when your application runs outside ROS 2.
Multi-Redundant MEMS with Advanced Skew-Redundant Geometry Mounting. olixAI™ combines calibration, filtering and sensing into one output.
02
Publish through olixOS™
Configure motion streams, coordinate frames and publication settings over the Ethernet connection.
03
Integrate in your robot
Receive standard ROS 2 topics or timestamped JSON in your estimator, logger or application.
Power and Ethernet data through one cable. Accessory configuration follows the quotation.
X2 UltraMotion sensing
PoE networkRobot host via Ethernet
Ethernet data + isolated PoE · one physical connection
Ethernet + isolated PoE
One cable. A direct connection.
Connect X2 Ultra to an IEEE 802.3af PoE switch or injector. Power the sensor, reach olixOS™ in your browser and stream motion data over the same Ethernet connection.
100 Mb/s EthernetMotion data, configuration and diagnostics on your robot’s network.
Isolated PoEIEEE 802.3af input, 37–57 V DC. End-span and mid-span power-feed support.
Defined installationApproved M12-to-RJ45 mating cable, with strain relief and clearance for the coupling and bend radius.
Ethernet + PoE
The connection, considered.
Your cable. Your configuration.
A simple connection starts with the right fit. Specify the cable length and connector arrangement for your robot.
01 Cable length
From the sensor to your PoE source.
02 Connector arrangement
Approved M12-to-RJ45 mating cable.
03 Installation
Room for the coupling, cable bend and strain relief.
Cable appearance and supplied configuration follow your quotation.
olixAI™ combines redundant sensing, calibration corrections and filtering into a unified motion output. olixOS™ brings configuration, calibration and diagnostics together on the device.
OrientationQuaternion x, y, z, w
Angular velocityRadians per second
Accelerationm/s² · gravity-compensated stream
Auxiliary dataMagnetic field, temperature, status
olixAI™ · Time-series sensor fusion
A sequence of motion. One coherent output.
Explore how successive inertial measurements can inform a unified motion state.
01 / Sensor sequences
Accelerationax · ay · az
Angular velocitygx · gy · gz
Magnetic fieldmx · my · mz
AHRS mode input
02 / Temporal fusion concept
olixAI™
Earlier samplesNow
03 / Unified motion
OrientationQuaternion · x, y, z, w
Angular velocityrad/s
Accelerationm/s²
One virtual measurement frame
IMU mode · relative orientation from acceleration and angular velocity. Magnetic-field data is excluded from this orientation solution.
Conceptual neural-fusion illustration. Waveforms, nodes and connections are illustrative; they do not disclose the internal network architecture or show measured data.
AI-enhanced sensor fusion
Keep the movement. Refine the signal.
Illustrative concept
See the fusion objective in a moving robot frame: follow changing orientation while limiting vibration and short disturbances. The traces below show the same scene as motion data.
Motion, made visible.Illustrative attitude response
Traditional fusion conceptIllustrated residual interference
olixAI™ conceptIllustrated refined motion output
Dashed outlines show the reference movement.
Follow the movement.
Conceptual relative-motion signal · the same scene, shown as data.
StartTime →
Changing orientation · preserve the movement needed by your state estimator.
Synthetic, normalized illustration of the fusion concept, not measured X2 Ultra outputs or a benchmark against a specific traditional algorithm. Robot frames and traces share one scripted motion sequence. Product specifications and their basis follow below.
Two modes. One virtual measurement frame. IMU mode provides relative orientation from acceleration and angular rate. AHRS mode adds magnetometer data for magnetic heading aiding. Fused output uses a virtual origin at the board centre. Calibrate the magnetic field in the installed configuration and apply the sensor-to-robot transform.
Fused motion output
Precision. In the details.
A closer look at orientation, noise and stability. Explore the X2 Ultra specification estimates.
Relative orientation · estimated
A precise view of movement.
<0.4°
Roll & pitch · IMU mode · estimated RMSE
Settled relative rotation
Conservatively rounded engineering estimate for aggregate relative-angle error after motion has stopped.
Gyroscope
Fused noise density
<0.01°/s/√Hz
Estimated bound · stationary three-axis mean
Accelerometer
Fused noise density
<200µg/√Hz
Estimated bound · stationary three-axis mean
Engineering estimates from the product datasheet, rounded conservatively. Model-specific performance validation is pending; these figures are not measured X2 acceptance limits or production guarantees.
Settled relative rotation. Estimated aggregate RMSE for settled single-axis rotations. This describes relative angular displacement after motion has stopped.
Moving platform. Estimated relative-angle error between settled poses on a moving platform. These values do not specify continuous tracking error or absolute magnetic-heading accuracy.
Reading the bounds. The < notation denotes an estimated aggregate bound, not a maximum sample error or a production guarantee. Performance depends on mode, calibration, temperature, range, bandwidth and installation. Compare configurations with matching filter and publication settings.
Estimated fused inertial noise and stability bounds
Parameter
Gyroscope
Accelerometer
Noise density
<0.01 °/s/√Hz
<200 µg/√Hz
Random walk
Angle: <0.6 °/√h
Velocity: <0.11 m/s/√h
Bias instability
<7 °/h
<0.001 m/s²
Engineering estimates for three-axis means during stationary operation at room temperature. Allan-derived white-noise coefficients and bias instability are distinct from absolute bias and thermal drift. Model-specific validation is pending.
Set up the sensor in your browser. Choose the output frame, profile, measurement range and stream settings, then bring the data into your application.
01 / CONNECT
Power and network
Mount the sensor on a rigid surface. Secure the approved Ethernet cable and connect its RJ45 end to a compatible IEEE 802.3af PoE switch or injector.
02 / CONFIGURE
Your sensor. Your frame.
Default device IP: 192.168.7.100. Example host IP: 192.168.7.1/24. Open the device address in a browser; configure mode, frame, publication rate and calibration.
03 / INTEGRATE
Receive motion data
Subscribe to ROS 2 topics with matching QoS, or enable JSON export. TCP uses newline-delimited samples; UDP begins after an initial client registration packet.
Use the commissioned device address and configured namespace. Place the host on the sensor subnet with a unique IP address. Match subscriber reliability to publisher QoS; give each sensor a unique IP address and ROS namespace.
Integrated velocity estimate; external aiding for sustained use
magneticfield
sensor_msgs/msg/MagneticField
Auxiliary magnetic field, in tesla
temperature
sensor_msgs/msg/Temperature
Internal temperature, in °C
status
diagnostic_msgs/msg/DiagnosticStatus
System diagnostics
Topic availability follows the selected mode; optional /tf publication is supported. TCP / UDP export uses the olive_imu.v1 JSON schema, with timestamps, frame ID, quaternion and motion data; default port 5600. The host integration reference uses ROS 2 Humble; match topic names and QoS to your configured device.
Designed to fit your system
Compact by design. Detailed where it matters.
A 48 × 48 mm housing footprint, integrated mounting base and a single network connector. Integrate X2 Ultra with a defined mechanical pose and virtual measurement reference.
Nominal dimensions in mm. Overall depth includes the 8.96 mm connector projection, rounded to 57 mm. Use the controlled drawing for production tolerances.
Mechanical & electrical
Mechanical and electrical specifications
Overall envelope
66 × 57 × 42 mm, including mounting tabs and sensor-side connector
Housing footprint
48 × 48 mm
Mounting base
66 × 48 × 4 mm
Mounting holes
4 × Ø4.1 mm; 58 × 28 mm centres
Network
100 Mb/s Ethernet
Power input
IEEE 802.3af PoE · 37–57 V DC
PoE power stage
12.95 W capacity; 1.5 kV input-to-output isolation
Installation
Approved mating cable; provide connector and cable-bend clearance
Sensing & configuration
Sensing and configuration specifications
Acceleration ranges
±2 / ±4 / ±8 / ±16 g in IMU mode; ±4 g in AHRS
Angular-rate front end
±125 / ±250 / ±500 / ±1,000 / ±2,000 °/s; available settings depend on configuration
Publication setting
0–1,000 Hz; 100 Hz default¹
Output conventions
FLU / FRD
Application profiles
Drone / Mobile Robot / Humanoid / Raw
Platform
olixOS™ with olixAI™
Processing resources
8-core 64-bit CPU, up to 2.2 GHz; up to 6 TOPS peak²
Memory / storage
4 GB memory / 64 GB eMMC, nominal
Ambient operation
−20 to +60 °C design envelope³
¹ Publication rate is a configurable setting; effective throughput depends on configuration and network. Measurement bandwidth and publication rate are separate. ² Peak platform capability; sustained performance depends on workload and thermal conditions. ³ Subsystem-based environmental design envelope; complete-assembly qualification is pending. Storage: −20 to +70 °C; humidity: 10–90% RH, non-condensing, on the same basis. PoE capacity and isolation ratings apply to the power stage; capacity is distinct from device consumption. Confirm assembled mass, measured operating power, connection drawing and applicable conformity documentation with the quotation.
Applications · Physical AI
Different robots. A common sense of motion.
From a robot’s next step to a vehicle’s next turn. Bring orientation, angular velocity and acceleration into the systems that move.
Humanoid robotics
Every step. In perspective.
Give your state estimator a body-motion reference. Combine X2 Ultra’s fused orientation, angular velocity and acceleration with joint encoders, contact information and external perception.
Humanoid profileBody orientationROS 2 integration
Define the body frame.Mount on a rigid torso or pelvis reference and apply the sensor-to-robot transform.
Configure the response.Select the Humanoid profile, output convention and publication settings in olixOS™.
Build the estimator.Combine inertial observations with the robot’s other measurements for gait and balance estimation.
Explore the applications
Find your next move.
Explore how an inertial reference fits into different robotic platforms.
Autonomous mobile robots
Motion for the next turn.
Bring orientation and inertial data into localization alongside wheel odometry and vision.
Automated forklifts
A reference for vehicle motion.
Add body-motion observations to a warehouse vehicle’s localization and navigation stack.
Legged robots
Understand body attitude.
Provide body orientation and angular rate to the estimator alongside joint and contact measurements.
Construction equipment
Measure platform movement.
Observe chassis orientation and angular motion during equipment development and integration.
Industrial tow trains
Follow the platform.
Combine inertial observations with vehicle odometry along changing factory routes.
Agricultural robotics
Motion across changing terrain.
Combine body-motion measurements with GNSS, vision and vehicle odometry in field-robot development.
Autonomous shuttles
An inertial view of the route.
Provide timestamped body-motion data to a vehicle’s externally aided state estimator.
Heavy mobile equipment
Observe chassis attitude.
Explore orientation and vibration observations on large mobile platforms.
Service robots
A consistent sense of motion.
Integrate orientation and inertial data with perception and localization in indoor mobile systems.
Mobile manipulation
Connect base and robot frames.
Observe mobile-base motion and maintain a defined sensor-to-robot installation transform.
Application illustrations show integration concepts, not customer deployments or certified use cases. Platform suitability, mounting and environmental requirements must be assessed for the complete installation. Inertial data complements the robot’s other sensors; it does not by itself provide position, safety functions or closed-loop balance control.
Integration, explained
The details. Made clear.
What is the difference between IMU and AHRS mode?
IMU mode provides relative orientation using inertial acceleration and angular-rate measurements; magnetometer data is excluded from that orientation solution. AHRS mode includes magnetometer data for magnetic heading aiding. Local orientation zero is configurable in both modes.
For AHRS operation, perform magnetic calibration in the installed configuration and maintain separation from magnets, high-current wiring and ferromagnetic structures. The relative-angle specifications do not define absolute magnetic-heading accuracy.
Does X2 Ultra publish at 1,000 Hz?
The main IMU publish-rate setting extends to 1,000 Hz, with a 100 Hz default. Effective throughput depends on configuration and network. Verify the received rate in your application and select compatible filter settings. Publication rate and measurement bandwidth are configured separately.
Where is the measurement origin?
Fused and filtered data is published at a virtual origin at the board centre, aligned with the central inertial reference frame. Use the configured output convention and apply the installation transform from this sensor frame to your robot frame.
How are bias calibration and orientation zero different?
Bias calibration estimates sensor bias while the device is stationary in its normal mounting pose. Run the olixOS™ bias action or setBias service. Orientation zero uses setZeroQuaternion to define the present orientation as the local reference.
For AHRS mode, magnetic calibration is a separate operation. Choose FLU or FRD consistently and confirm the active settings after any requested service restart.
Can I use the sensor without ROS 2?
Yes. Enable TCP or UDP JSON export in olixOS™. The sensor acts as the server, using a configurable port with 5600 as the default. TCP provides newline-delimited JSON samples; UDP requires an initial client registration packet. Both include timestamps and the configured frame ID.
How do I integrate multiple X2 Ultra sensors?
Connect the sensors to a compatible PoE network. Assign each device a unique IP address and ROS namespace, and confirm that the switch or injector has the per-port capacity and total power budget for your configuration. Use a shared time reference when correlating data from multiple devices.
Which cable and mounting accessories should I order?
Specify an approved M12-to-RJ45 Ethernet cable, its length and connector arrangement. Connect to an IEEE 802.3af PoE switch or injector. Use the order-specific connection drawing and provide strain relief and clearance for the coupling and cable bend.
The mounting base has four Ø4.1 mm holes on 58 × 28 mm centres. Confirm delivery contents, assembled mass and measured operating power with the quotation.
Documentation & next steps
Documentation. And direct support.
Technical reference
X2 Ultra datasheet
The complete technical reference. Performance, interfaces, installation drawings and integration details in one PDF.