Tabletop Pen Manipulation With a Vision-Guided 4-DoF Arm
This paper demonstrates that a low-cost, underactuated 4-DoF robotic arm can successfully pick up and color-sort writing utensils at arbitrary orientations by leveraging a vision-guided strategy that combines YOLO11n-OBB detection with corrective sweeping motions to compensate for the lack of a wrist rotation joint.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Robots have long been the domain of heavy industry, where massive, expensive machines perform repetitive tasks with perfect precision. But for students, small labs, and independent researchers, the dream of a versatile robotic helper has often been blocked by cost and complexity. A robot arm that can reach in any direction and twist its "wrist" to grab an object from any angle typically requires six moving joints, or degrees of freedom. Each extra joint adds significant expense, pushing the price of such machines into the tens of thousands of dollars. This leaves a gap for those who need a robot but cannot afford a full industrial cell. The question becomes whether a simpler, cheaper machine with fewer moving parts can still do useful work, or if it is fundamentally limited by its lack of flexibility.
In a recent study, researchers Anirudh Rangarajan and Bibit Bianchini explored this exact challenge using a low-cost robotic arm that costs roughly $200. This device, the Waveshare RoArm-M2-S, has only four moving joints. It can move up, down, left, right, and forward, and it can open and close its gripper. However, it lacks the fifth joint needed to rotate the gripper sideways, a movement usually required to align with an object lying at a strange angle on a table. Theoretically, this missing joint should make the robot useless for picking up items like pens that are scattered in random directions. To solve this, the team did not try to build a better robot; instead, they built a smarter way to see and think. They equipped the arm with a camera and a computer program that could calculate how to move the pen into a position the robot could actually grab, effectively using software to compensate for the missing hardware.
The experiment took place on a simple tabletop setup. A camera was mounted directly above the workspace, looking down at a row of four bins colored blue, red, green, and gray. Scattered on the table were various writing utensils, including pens and markers, lying at different angles. The goal was for the robot to identify each pen by its color and sort it into the correct bin. Because the robot could not twist its wrist to match the pen's angle, it had to rely on a specific strategy. The computer first used a visual detection system to find every pen and measure its orientation. If a pen was lying nearly straight toward the robot, the arm would simply reach down and grab it. But if the pen was angled sharply away, the robot would not try to force a grab. Instead, it would perform a gentle, sweeping motion, nudging the pen across the table until it rotated into a position where the robot could pick it up.
The researchers tested this system with seven different writing utensils, varying their colors and shapes. They ran hundreds of trials, manually triggering the robot to pick up one pen at a time and observing how it handled different starting positions. The system proved remarkably effective. Out of 326 total movements recorded, the robot successfully grabbed 196 pens directly without any adjustment. For the remaining 130 pens, which were lying at steeper angles, the robot performed a corrective sweep to reorient them. These sweeping motions were not random; the computer calculated exactly how far to push the pen and in which direction to rotate it. In many cases, a single sweep was enough to turn a difficult angle into an easy one, allowing the robot to complete the task. The system also correctly identified the color of the pens in the vast majority of cases, sorting them into the appropriate bins.
The success of this project suggests that clever engineering can overcome hardware limitations. By combining a precise visual system with a motion plan that adapts to what it sees, a robot with fewer moving parts can perform tasks that usually require more expensive, complex machines. The researchers found that the robot could correct misalignments of up to 90 degrees, proving that the missing joint was not a fatal flaw but merely a constraint that could be worked around. While the system did encounter some challenges, such as difficulty with pens lying at extreme angles or under very bright glare, these were isolated issues that did not prevent the overall success of the task. The study demonstrates that for structured tasks like sorting objects on a table, the intelligence of the software can be just as important as the physical capabilities of the machine.
This work highlights a shift in how we might approach robotics in the future. Instead of assuming that a robot needs more joints to be more capable, we can design systems that use perception and planning to achieve the same result with simpler, more affordable hardware. The researchers showed that a $200 arm, when guided by a camera and a smart algorithm, can reliably pick up and sort objects that are scattered in random orientations. This approach makes advanced manipulation tasks accessible to a wider range of users, from schools to small startups, without the need for massive budgets. The findings suggest that the path to more capable robots may not always lie in building bigger machines, but in teaching smaller ones how to think.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.