ROBOTICS & AVs
Manipulation & Grasping
Manipulation & Grasping. What it takes to run it.
Precise 3D object pose estimation for robotic arms.
// 01 · The question
“Where do I grip this, and in which orientation?”
// 02 · Why it is hard
Objects arrive in bins, overlapping and at arbitrary angles, often shiny or transparent. Pose has to be inferred from a partial view of a thing that is partly buried under other things.
// 03 · What the labels have to be
Keypoints and 6-DoF pose
Decided before the first frame
A gripper needs a position and a rotation in space, and the graspable features are specific points on the object: a handle, a rim, a lip. Neither a box nor a mask carries orientation, and orientation is most of the answer.
This is the decision that is expensive to reverse. The geometry has to match what the answer contains, and finding out it does not means labelling the set a second time.
// 04 · How we run it
Three parts of one loop, on this job.
LexAnnotate
Pixel-Perfect Data
Pixel-level semantic segmentation, LiDAR point cloud annotation, sensor fusion labeling, and temporal event tracking for complex perception tasks. Expert annotators, accountable for every label.
LexInsight
Spatial Intelligence
Build spatial understanding from flat labels. Map object relationships, track movement patterns, identify zero-day anomalies that standard training data misses.
LexAlert
Perception Alerts
Continuously validate perception performance in production. Detect when sensor drift or environmental changes degrade navigation. Route failures back for retraining before they cause incidents.
// 05 · What breaks it after launch
A new gripper or a new supplier's version of the same part changes what a valid grasp is. The perception is unchanged and the grasps it proposes stop working, which reads as a perception failure and is not one.
It is not the only one that can get this use case, it is the one that usually gets it first. All ten conditions.
// 06 · In robotics
Dynamic Spatial Navigation
The Challenge
Flat bounding boxes failed to provide spatial context for autonomous material handling in shifting environments.
Our Execution
Deployed pixel-level semantic segmentation and temporal event tracking for high-fidelity spatial datasets.
The Result
Enabled autonomous systems to adapt to changing environments safely. Reduced manual overrides.
95%+
Navigation reliability in changing environments