Robotics / Edge AI / 2026
Autonomous Pantry Robot
Developing a pantry-organizing robot concept in Onshape. I started with the end effector, then plan to build the arm, perception, and local-compute layout around it.
Onshape end-effector CAD for a pantry robot concept built around shelf-scale manipulation.
- Role
- Mechanical concept, CAD, system architecture
- Status
- In progress
- Tools
- OnshapeComputer visionCoral NPUEnd effectorGripper assemblyShelf constraintsLocal inference

Problem
Design a compact robot concept for recognizing pantry items, reaching into shelf spaces, and moving objects with local perception.
Current work
- Started the CAD work with the end effector, using the hand assembly as the first piece of the manipulation system.
- Defined the near-term mechanical questions: shelf clearance, grip stability, wrist stiffness, cable routing, and how the end effector will mount to a larger arm.
- Mapped the next pieces of the robot around local computer vision, edge inference, arm packaging, and a compact base.
Engineering decisions
- Treat the current CAD as an end-effector development step, then grow the robot from a subsystem that already has geometry.
- Use the gripper geometry as a reference while I work through shelf clearance, mounting, validation, and system packaging.
- Plan the local-AI direction around Google Coral NPU-class edge hardware and compact Gemma models.
- Keep the AI hardware and mechanical packaging as separate design questions until both are easier to test.
Current state
- The project currently centers on the end-effector CAD assembly and the pantry-specific constraints around it.
- The next design step is building the arm and base around this hand geometry.