SO100 Robot Arm Projects — 10 Cool Things You Can Build in 2026

Looking for SO100 robot arm projects or wondering what you can do with a robot arm once it arrives? These 10 ideas take you from a simple robot arm pick and place demo to serious LeRobot projects with vision, imitation learning, and multi-arm coordination.

---

The best thing about the SO100 is that it does not lock you into one narrow use case. You can start with a quick teleoperation demo on Friday night, build a sorting system on Saturday, and by Sunday be recording demonstrations for an imitation learning policy. That is exactly why the arm has become such a strong platform for LeRobot projects, classroom experiments, and maker builds in 2026.

If you are still getting your hardware online, start with our LeRobot + SO100 setup tutorial. If you want the broader hardware context first, read the SO100 complete guide. Once the arm is powered, calibrated, and moving, the project ideas below are the fastest way to turn it into something that feels real.

Here are 10 cool robot arm projects for 2026, ranked to give you a mix of quick wins, portfolio-worthy builds, and advanced experiments that justify owning the hardware.

---

1. Pick-and-place sorting by color or shape

Difficulty: 1/5
Best for: First weekend build, classroom demos, benchmarking

This is the classic starting point because it teaches the foundations of almost every other manipulation workflow. Set up a few objects on your desk, create two or three bins, and teach the arm to move pieces based on a simple rule such as color, shape, or size.

Why it works so well:

• You get fast visual feedback when the robot succeeds or fails
• The task scales easily from scripted motion to AI control
• It maps directly to real warehouse and manufacturing use cases

Start simple with wooden blocks, LEGO bricks, or bottle caps. Once your baseline works, add clutter, randomize object positions, or narrow the target bins so the project becomes a real precision test instead of a canned demo.

If you want a clean path from this project into learning-based control, pair it with the imitation learning guide. Sorting is one of the most natural gateways into data collection.

2. Teleoperation with a leader arm

Difficulty: 1/5
Best for: New owners, demos, proving the hardware works

Teleoperation is the moment when the SO100 stops feeling like a box of parts and starts feeling like a robotics platform. Move the leader arm by hand and watch the follower mirror it in real time. That alone is compelling, but it is also the control method that unlocks everything else in the stack.

This project is worth doing even if it sounds basic because it helps you verify:

• Servo direction and calibration
• USB and power stability
• Workspace layout
• Smoothness at different motion speeds

Once teleoperation is stable, you can record demonstrations, test reach limits, and build repeatable tasks around motions that already feel natural. It is the best first answer to the question, "What can you do with a robot arm right away?"

3. Imitation learning — teach by demonstration

Difficulty: 3/5
Best for: AI builders, students, portfolio projects

This is the flagship SO100 workflow. Instead of writing every motion by hand, you demonstrate a task with the leader arm, record multiple episodes, and train a policy that can replay the behavior autonomously on the follower arm.

A good first imitation learning task should be:

• Short, ideally under 10 seconds
• Highly repeatable
• Easy to reset between episodes
• Visually obvious when it succeeds

Pick-and-place, block insertion, and object sorting are strong starting points. If you want to move beyond "robotics demo" into "AI robotics project," this is the project that matters most. Our imitation learning with robot arms guide explains how demonstrations, datasets, and policies fit together, while the setup tutorial covers the practical LeRobot workflow.

4. Drawing or writing with a pen attachment

Difficulty: 2/5
Best for: Social content, calibration practice, classroom engagement

Attach a marker or pen to the end effector and use the SO100 to draw lines, geometric patterns, signatures, or short words. This project looks playful, but it teaches real lessons about repeatability, surface alignment, and motion smoothing.

A few ways to level it up:

• Start with straight lines and circles to tune offsets
• Trace printed templates taped to the desk
• Write letters or initials at different scales
• Compare teleoperated drawing versus prerecorded motion playback

Because the output stays on the page, drawing makes small positioning errors obvious. It is one of the fastest ways to improve your setup discipline and create a project that is easy to share with other people.

5. Chess or checkers playing robot

Difficulty: 3/5
Best for: Makers, streamable demos, board-game automation

A tabletop game robot is one of the most recognizable robot arm project ideas because everyone instantly understands the objective. You do not need a full grandmaster chess engine to make it interesting. Even a checkers bot that can pick up pieces cleanly and move them to legal squares is already a strong demonstration of planning plus manipulation.

The practical version of this project usually looks like this:

• Use high-contrast pieces that are easy to grasp
• Mark the board so each square has a consistent coordinate
• Start with scripted legal moves before adding vision
• Upgrade later to camera-based piece detection and move planning

This project forces you to think about approach angle, collision avoidance, and where to place the gripper so pieces do not tip over. It also creates a natural bridge into camera calibration if you want the robot to identify the board state visually.

6. Automated lab sample handling

Difficulty: 4/5
Best for: Research prototypes, lab training, repetitive transfer tasks

If you want the SO100 to feel less like a desk toy and more like a serious automation platform, build a sample-handling routine. Use empty tubes, capped vials, or color-coded mock containers and move them between racks, trays, or stations.

This is a strong project because it combines:

• Careful grasping
• Consistent placement
• Repeatable sequencing
• Workspace organization

Use mock samples first. The goal here is education and prototyping, not regulated clinical work. Even without real wet-lab use, the project gives you a realistic automation pattern that looks much closer to real research infrastructure than generic block moving.

If you are designing a larger maker or classroom workflow, our how to build a robot arm with AI guide is a useful companion for the tooling and environment side.

7. Object detection plus grasping with a camera

Difficulty: 4/5
Best for: Computer vision workflows, autonomous demos, advanced LeRobot projects

This is where your SO100 starts reacting to the world instead of just replaying positions. Mount a webcam above the workspace, run object detection or simple segmentation, and use the camera feed to decide what to pick up next.

A solid version of this project includes:

• A fixed camera angle so calibration stays stable
• High-contrast objects at first
• A small set of known classes such as cubes, cups, and markers
• Simple grasp heuristics before end-to-end learning

This project is one of the best answers to "What can you do with a robot arm that feels genuinely intelligent?" Once the arm can detect an object, move to it, and grasp it without you placing it in exactly the same spot every time, the leap in capability is obvious.

8. Dance or gesture choreography

Difficulty: 2/5
Best for: Events, content, timing and motion experiments

Not every project needs to be utilitarian. Dance and gesture choreography are perfect if you want a fast, memorable build that shows the arm can move expressively. Create short routines, wave patterns, synchronized gestures, or music-timed motion clips.

Why it is worth building:

• It is quick to demo at meetups or in videos
• You learn timing, repeatability, and motion pacing
• It helps expose vibration or jerkiness in your trajectories

If you have two arms available, mirrored choreography becomes even better. This is the easiest path into multi-arm timing before you attempt more demanding handoff or collaborative manipulation tasks.

9. Pouring liquids with precision control

Difficulty: 4/5
Best for: Smooth trajectory control, benchmark tasks, advanced demos

Pouring is a deceptively hard robot arm project. It requires precise approach, controlled tilt, and a steady return motion so the container does not slosh or spill. That makes it one of the best ways to stress-test whether your control loop is smooth or just "good enough" for block-moving clips.

For the first version, use:

• Water with food coloring, or dry materials like rice
• Small cups that fit securely in the gripper
• A wide target container before moving to narrow openings
• Measured fill targets so you can compare runs

Once you can pour consistently, you can experiment with speed versus accuracy, container shapes, and imitation learning on the full sequence. It is one of the most satisfying advanced tasks because the result is easy to judge without specialized tools.

10. Multi-arm collaboration with 2x SO100

Difficulty: 5/5
Best for: Research-style builds, ambitious makers, standout demos

If you want a project that immediately separates your setup from the average hobby robot arm desk, run two SO100 systems together. That gives you a platform for handoff tasks, staged assembly, mirrored motion, and simple bimanual coordination.

A realistic progression looks like this:

1. Run two independent pick-and-place routines in parallel
2. Add timing so one arm waits for the other at a transfer point
3. Pass an object from one workspace to the other
4. Experiment with synchronized behaviors or shared camera input

This is the most advanced build on the list because it compounds every challenge: calibration, timing, collision risk, workspace planning, and software orchestration. It is also one of the coolest robot arm projects you can show in 2026 because it immediately communicates scale and ambition.

---

What you need to get started

You do not need a massive lab to build these projects. Most people can get started on a desk with a simple bill of materials:

Item	Why you need it	Notes
SO100 Complete Kit	Core hardware for every project	Includes pre-assembled leader + follower arms, 12 STS3215 servos, 2 driver boards, USB-C cables, and power supply
Computer with Python 3.10+	Runs setup, control, and LeRobot	Linux is ideal, but macOS and Windows can work
LeRobot	Teleoperation, recording, training, deployment	Best choice for AI-focused SO100 projects
Small objects	For pick-and-place, sorting, and grasping	Blocks, cups, caps, game pieces, sample tubes
Webcam	Required for vision-guided grasping	Any stable 720p+ USB camera is enough to start
Pen holder or marker mount	For drawing and writing	Can be improvised at first
Good lighting	Helps both demos and camera-based tasks	Especially useful for detection and board games
Second SO100 setup	Needed for multi-arm collaboration	Optional, but powerful for advanced builds

The simplest path is to buy a complete kit, follow the LeRobot setup tutorial, and start with teleoperation plus sorting before moving into vision or imitation learning. If you are still deciding between prebuilt and DIY, the buying guide and build guide will save you time.

Where to buy

If you already know you want to build one of these projects, the fastest option is the SO100 Complete Kit. It removes the slow part of the process: hunting for parts, printing components, and debugging assembly before you ever touch the software.

Buy the SO100 Complete Kit for $199 →

That gets you the leader + follower setup needed for teleoperation and imitation learning, plus the core electronics needed for pick-and-place, drawing, grasping, and most LeRobot projects on this list. If you want a seller comparison first, read where to buy the SO100 robot arm.

FAQ

What can you do with an SO100 robot arm?

You can build sorting systems, teleoperation demos, imitation learning workflows, drawing bots, board-game robots, sample handling prototypes, vision-guided grasping, pouring demos, and multi-arm collaboration setups. The arm is flexible enough for both quick maker projects and more serious AI robotics experiments.

Which SO100 project should beginners start with?

Start with teleoperation and a simple robot arm pick-and-place task. Those projects verify your hardware, teach basic workspace setup, and give you a clean starting point for sorting, drawing, and imitation learning later.

Do I need a camera for LeRobot projects?

No. Many great SO100 robot arm projects start without vision. Teleoperation, drawing, chess motion playback, and basic sorting all work without a camera. Add a webcam when you want object detection, position generalization, or camera-conditioned policies.

Is the SO100 good for portfolio projects?

Yes. A well-documented SO100 build shows hardware integration, Python workflows, calibration, control logic, and optionally AI training. That combination is much stronger than a purely simulated project if you want to demonstrate practical robotics skill.

Should I buy one SO100 or two?

One complete kit is enough for almost every project on this page because it already gives you the leader + follower setup. Buy a second setup when you are ready for multi-arm coordination, handoff workflows, or more ambitious research-style demos.

---

The best project is the one you can finish quickly enough to learn from. Start with a simple success, publish the video, then iterate. That is how most strong SO100 builders go from "I just unpacked this thing" to a stack of real robot arm projects in a few weekends.