01 / HUMANOID ROBOTS
Whole-body state
Torso and pelvis attitude, limb dynamics, balance recovery, fall detection and synchronized segment motion.
Inertial Motion Intelligence
Every move.
Understood.
Olive combines patented Multi-Redundant MEMS technology, AI-based sensor fusion and source-accurate timing to deliver fast, accurate and low-latency inertial state directly to the robot.
Patented Multi-Redundant MEMSAI-based sensor fusionUp to 1000 Hz<1 ms low latency
Source time / PTP locked
Inertial intelligence / active
The inner feedback loop
Commanded motion is not
measured motion.
Motors express intent. The IMU measures what the body actually did—rotation, acceleration, impact, tilt and vibration—so the estimator and controller can close the loop around reality.
Closed-loop robotics / physical state feedbackState observable
Balance / body attitudeHigh-rate inertial state feeds stabilization and recovery around the robot’s actual motion.
Gravity vectorbody relative to vertical
Angular ratefast rotational response
Confidencestate quality is observable
Observe body motionMeasure fast rotational and translational dynamics.
Bridge perception gapsMaintain a high-rate inertial reference between external updates.
Stabilize controlFeed balance, attitude and motion controllers with current state.
Detect disturbanceExpose shock, vibration, slip and unexpected motion.
From sensing to inertial intelligence
The IMU feels motion.
The AHRS understands it.
Accelerometers and gyroscopes observe specific force and angular rate. An AHRS combines them—optionally with magnetic heading and other references—to estimate a usable attitude state.
Motion representation / live layerAHRS estimate
+X+Y+Z
FUSED ATTITUDE / BODY FRAME
Estimates orientation and heading
Sensor fusion tracks a quaternion or roll, pitch and yaw while estimating bias, gravity and uncertainty. Magnetic heading remains environment-dependent.
Quaternionorientation without singularities
Gravity vectorbody relative to vertical
Heading + covariancereference and confidence
Multi-Redundant MEMS · Patented technology
One motion.
Multiple independent observations.
Olive’s patented Multi-Redundant MEMS architecture observes the same physical event across parallel inertial channels. olixAI compares residuals, isolates inconsistent data and publishes an accurate, confidence-aware motion state.
X2 / multi-redundant observation fieldMulti-channel nominal
olixAI fusionConsensus activealignment · residuals
outlier gating · confidence
bias + thermal compensation
outlier gating · confidence
bias + thermal compensation
REDUNDANT MEMSGYRO + ACCEL / VALID
REDUNDANT MEMSGYRO + ACCEL / VALID
REDUNDANT MEMSGYRO + ACCEL / VALID
REDUNDANT MEMSGYRO + ACCEL / VALID
REDUNDANT MEMSGYRO + ACCEL / VALID
Residual agreement / nominalParallel channels remain inside the active consensus envelope.
IN BAND
IN BAND
IN BAND
IN BAND
IN BAND
REDUNDANT OBSERVATIONMultiple channels observe the same physical event.
RESIDUAL ANALYSISChannel disagreement becomes a measurable diagnostic.
OUTLIER GATINGInconsistent data can be down-weighted or excluded.
CONFIDENCE OUTPUTState quality travels with the estimated motion.
olixAI™ Inertial Motion Intelligence
Raw motion in.
Decision-ready intelligence out.
olixAI performs adaptive, AI-based sensor fusion directly at the sensor edge—combining inertial observations, temporal context and diagnostics into a fast, accurate and confidence-aware state with minimal transport delay.
olixAI / VIBRATIONEdge inference
Raw observationFused state
Representative visualizationVibration / structured disturbanceThe local model separates motion context from machine-coupled vibration before publication.
Residual modelcontext-aware correction
State preservedphysical trend remains visible
Confidence trackedquality travels with output
01 / ACQUIRESource-timestamped Multi-Redundant MEMS observations.
02 / COMPENSATEBias, thermal and disturbance-aware preprocessing.
03 / FUSE + INFERState estimation, temporal features and confidence.
04 / PUBLISHRaw, fused and diagnostic ROS 2/DDS outputs.
Latency + source time
The right value at the wrong time
is the wrong state.
Fast dynamics require both a high update rate and a trustworthy timestamp. X2 processes motion locally, attaches time at the source and aligns distributed nodes through hardware-assisted IEEE 1588 PTP.
Distributed motion time domainPTP locked
1000 Hzup to motion output rate
<1 msmotion-to-message latency
IEEE 1588hardware-assisted PTP
SampleMEMS acquisition
Source timestampphysical event time
Edge fusionstate + covariance
DDS publishrobot graph
System note: end-to-end timing depends on the PTP grandmaster, switches, topology, traffic, host scheduling and DDS configuration. Validate the complete time domain under representative load.
Built for every kind of motion
The inner sense of
every moving robot.
Wherever stability, orientation, motion phase or dynamic response matters, inertial state becomes a foundational input to estimation and control.
02 / LEGGED ROBOTS
Dynamic balance
Base attitude, gait transition, impact response, foot-slip cues and high-rate state during aggressive locomotion.
03 / EXOSKELETONS & WEARABLES
Segment kinematics
Gait phase, joint-chain orientation and repeatable correlation with force plates or clinical references.
04 / AMR & AGV
Slip-aware navigation
Heading, tilt, vibration and inertial continuity for localization, odometry fusion and payload-aware motion.
05 / UAV & AERIAL
Attitude stabilization
High-rate angular motion, vibration-aware fusion and synchronized state for flight control and navigation.
06 / INDUSTRIAL MACHINERY
Dynamic machine state
Alignment, shock, resonance, tool motion and vibration signatures for control and condition monitoring.
Robot-native from first power-on
From physical motion
to the ROS graph.
X2 publishes raw, fused and diagnostic motion information through native ROS 2/DDS—with source time, covariance and device health preserved as part of the sensing contract.
X2 motion node / data contractPublishing
01 / ACQUIREMulti-Redundant MEMSacceleration · angular rate
magnetic field · temperature
magnetic field · temperature
02 / UNDERSTANDolixAI + AHRSorientation · bias · covariance
confidence · diagnostics
confidence · diagnostics
03 / PUBLISHROS 2 + DDSsource time · QoS · Web UI
OTA · Zenoh · TCP · UDP
OTA · Zenoh · TCP · UDP
REPRESENTATIVE MOTION CONTRACT
topic: /imu/data type: sensor_msgs/msg/Imu stamp: source_time / ptp_domain_0 outputs: orientation angular_velocity linear_acceleration covariance temperature status + confidence
Inertial Motion Intelligence · engineered at the source
Fast enough to control.
Accurate enough to trust.
Build with patented Multi-Redundant MEMS technology, olixAI sensor fusion and synchronized robot-native data—delivering low-latency motion intelligence from the first sample onward.
Integration responsibility: humanoid, wearable, rehabilitation, human-rated and safety-related applications require system-level hazard analysis, validation and applicable certification by the integrator.
