Wheelchair Assistive Arm With Shared Autonomy

ISEF Category: Robotics and Intelligent Machines

Subcategory: Other  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

A robot arm can make a cup of water feel close again. That is the promise of assistive robotics, and it matters most when a small mistake turns a simple reach into a hard task. Your project can measure whether smart control helps people finish a grasp faster and with less effort.

What Is It?

This project looks at a wheelchair-mounted robotic arm that helps a user grab nearby objects. The arm has 5 degrees of freedom, which means it can move in five independent ways, like a shoulder, elbow, and wrist doing different jobs. Instead of making the user control every tiny motion, shared autonomy gives the robot some of the work. It can also “snap” toward nearby graspable objects, which means the software guesses what you want and nudges the arm toward that target.

Think of it like autocomplete for motion. You still choose the goal, but the robot fills in the easiest path. In this project, you would compare one control style, like sip-and-puff or a head-IMU, with a joystick-only setup. A sip-and-puff controller reads pressure from breathing. A head-IMU reads head motion with an inertial measurement unit. Your main outcome can be task-completion time, but you can also track misses, retries, or path smoothness.

Why This Is a Good Topic

This is a strong science fair topic because you can test a clear question with real numbers. You can compare control methods, measure time, and look at how much help shared autonomy adds. The project connects to accessibility, human-robot interaction, and daily living aids for people with limited hand function. You can learn robot control, experimental design, and basic statistics while building something that solves a real need.

Research Questions

• How does shared autonomy affect task-completion time compared with joystick-only control?
• What is the effect of sip-and-puff control versus head-IMU control on reach accuracy?
• Does snap-to-object assistance reduce the number of failed grasps during pickup tasks?
• To what extent does object size change the benefit of shared autonomy?
• Which control method produces the smoothest end-effector path during a tabletop reach task?
• How does target distance affect completion time under assisted and unassisted control?

Basic Materials

• Wheelchair-mounted robotic arm prototype or benchtop 5-DOF arm.
• Control input device, such as a joystick, sip-and-puff switch, or head-IMU.
• Laptop or microcontroller for control logic.
• Camera or phone tripod for task timing and video review.
• Small graspable objects of different shapes and sizes.
• Marked tabletop workspace with fixed starting and target zones.
• Safety stop switch or emergency cutoff.
• Tape measure or printed grid for workspace layout.

Advanced Materials

• 5-DOF robotic arm with servo or motor encoders.
• Head-IMU sensor module or sip-and-puff interface hardware.
• Microcontroller or embedded computer for motion control.
• Force sensor or gripper current sensing for grasp feedback.
• Motion capture system or overhead video setup for path analysis.
• Wheelchair mount or adjustable test rig.
• Calibration targets with known geometry.
• Data acquisition software for logging commands, pose, and timing.

Software & Tools

• ImageJ: Measures task timing, path overlays, and movement traces from video frames.
• Python: Cleans data, computes summary statistics, and plots control comparisons.
• OpenCV: Tracks the robot end effector and object position in video.
• R: Runs statistical tests and makes publication-style figures.
• MIT OpenCourseWare: Offers free robotics and control lectures that help you understand shared autonomy.

Experiment Steps

1. Define one task, such as reaching, aligning, and grasping a small object from a fixed workspace.
2. Choose the control modes you will compare and decide what assistance level counts as shared autonomy.
3. Build a repeatable test setup that keeps object location, lighting, and starting pose consistent.
4. Plan your outcome measures, such as completion time, success rate, path length, and number of correction attempts.
5. Design controls that separate input-device effects from autonomy effects, so each trial answers one question.
6. Choose a data analysis plan before testing, so you know which comparisons and graphs will support your claim.

Common Pitfalls

• Letting the arm start from a slightly different pose each trial, which makes timing differences hard to trust.
• Testing joystick control and assisted control in different lighting or camera conditions, which changes perception and tracking quality.
• Choosing objects that are too easy to grasp, which hides any benefit from shared autonomy.
• Measuring only completion time and ignoring failed grasps, which can make a slow but reliable mode look worse than it is.
• Changing the assistance logic while collecting data, which mixes two experiments into one.

What Makes This Competitive

A stronger version of this project goes beyond simple speed tests. You can compare multiple users, multiple object types, and more than one assistance rule, then analyze error rates, path smoothness, and learning effects. If you add a clear baseline, strong controls, and a careful statistical test, your results become much more persuasive. A good project here answers not just whether assistance helps, but when, why, and for whom.

Project Variations

• Test the same control system on cups, bottles, and utensils to see which object shapes benefit most from snapping assistance.
• Compare head-IMU control with sip-and-puff control for users who must complete the same pickup task.
• Measure how much shared autonomy helps when the target objects are closer together, which raises selection errors.

Learn More

• NASA NTRS: Search for papers on assistive robotics, shared autonomy, and human-robot interaction design.
• PubMed: Search for review articles on upper-limb assistive robots and wheelchair-mounted manipulators.
• IEEE Xplore: Find peer-reviewed papers on shared control, intent inference, and robotic grasp assistance.
• MIT OpenCourseWare: Look for free robotics and control systems courses that explain kinematics and feedback control.
• NIH 3D Print Exchange: Explore free design ideas for assistive device parts and mounting concepts.