Solutions
Up to this point, you’ve explored individual Skills across the Telekinesis ecosystem — from Cornea (segmentation) and Retina (detection) to Vitreous (point cloud processing) and beyond. But real-world robotics systems don’t use skills in isolation. They combine them into end-to-end pipelines that solve meaningful problems.
Here, you’ll find complete, application-driven pipelines that show how different Skills work together — across packages — to solve tasks such as:
- Performing 6D pose estimation
- Aligning 3D scans with CAD models
- Segmenting objects on conveyors
- Picking items from bins
- Depalletizing boxes
- Navigating warehouse floors
Each example demonstrates how multiple Skills can be orchestrated into a coherent workflow — from raw sensor input to actionable outputs.
These examples are adapted from the telekinesis-examples repository. repository. Before diving in, make sure you’ve installed the telekinesis-ai skill library by following the Quickstart guide.
Available Solutions
Each use case is tailored to a specific application domain so you can see how the same underlying Skills adapt across industries.
| Use case | Goal | Best for |
|---|---|---|
| Object pose estimation from RGB images | Estimate 2D poses (centroid, bbox) from a text prompt via Qwen + SAM | Grasping, tracking, input to 3D pose estimation |
| Object pose estimation from RGBD using bounding box | Estimate 3D poses (centroids) from a point cloud with a 3D bbox ROI and DBSCAN | Bin picking, multi-object 3D pose, ROI processing |
| Object pose estimation from RGBD using mask | Estimate 3D pose (centroid) from a point cloud filtered by a 2D mask | Single-object 3D pose, RGB+depth fusion, grasping |
| Planar pose estimation from RGB images | Estimate 6-DOF planar poses from one image (segment, contours, PCA, backproject) | Bin picking, part validation, single-camera planar pose |
| Conveyor segmentation from RGB images | Segment packages or items on a conveyor belt | Logistics, sorting, quality checks |
| Depalletizing box segmentation from RGB images | Segment boxes or cases on a pallet | Robotic depalletizing, unloading |
| Bin picking object segmentation from RGB images | Segment individual objects inside a bin | Robotic pick-and-place, kitting |
| Ground region segmentation from RGB images | Segment ground or floor for navigation | Mobile robots, AGVs, mapping |
| PCB region segmentation from RGB images | Segment PCB or component regions | Inspection, assembly, AOI |
| Pedestrian segmentation from RGB images | Segment pedestrians for safety and navigation | Safety zones, people counting, obstacle avoidance |
| Forklift segmentation from RGB images | Segment forklifts or industrial vehicles | Warehouse safety, fleet tracking, collision avoidance |