Building industrial robots do not necessarily mean spending lots of money on expensive hardware. Built with fischertechnik (a construction toy originating from Germany) and controlled by an Arduino Mega, this delta robot is a fairly simple machine. Nonetheless, it does not lack speed or precision to fulfill its job: The robot picks items from a conveyor and sorts them by color similarly to industrial pick and place robots.

What are the main challenges of this project?

• Delta robot kinematics
• Motion tracking of items on conveyor
• Vacuum gripping

Delta Robot Kinematics

Unlike simple kinematics such as linear kinematics that are often used in 3D printers, delta robot kinematics are more complicated and not straightforward to understand. To control a delta robot with cartesian coordinates, so-called inverse kinematics is required. I've implemented inverse kinematics for delta robots according to a paper published by Robert L. Williams II: "The Delta Parallel Robot: Kinematics Solutions"

Technical Deep-Dive: Delta-Parallel Inverse Kinematics

• The Williams II Algorithm: Unlike standard Cartesian or SCARA robots, a Delta robot's end-effector position is a non-linear function of its three upper-arm angles $(\theta_1, \theta_2, \theta_3)$. The project implements the Robert L. Williams II "Kinematics Solutions" algorithm, which performs high-speed trigonometric forensics to calculate motor steps for a desired $(x, y, z)$ waypoint. This allows for smooth, curvilinear paths necessary for industrial picking.
• Workspace Diagnostics: The parallel-linkage architecture ensures the moving load (the "bicep" and "forearm") remains minimal, allowing the robot to attain high operational frequencies without the centrifugal drift typical of serial manipulators.

Although this seems to be computationally expensive, delta kinematics have some good advantages over other robot kinematics. Firstly, they are astoundingly fast and precise, secondly they have quite a good range of operation. However, vibrations are a serious issue and moving load should be as low as possible.

The following video gives you an intuition of what is possible with delta robots:

Motion Tracking The position of every item is captured by a laser light barrier. This lets the robot know when the item has passed a certain position on the conveyor. Furthermore, the conveyor speed is measured by reading a encoder signal. This lets the robot know how fast the item is moving. Given that information, the robot's movement can be calculated so that it picks the item at the right time.

Technical Deep-Dive: Dynamic Conveyor Synchronization

• Laser-Gate & Encoder Fusion: The robot achieves "On-the-Fly" picking by fusing data from two sensing nodes. A laser light barrier detects the entry of an item $(\text{Time}{0})$, while a rotary encoder provides the instantaneous conveyor velocity $(\nu)$. The Arduino Mega calculates the intercept coordinate $(x{pick})$ based on the elapsed time, ensuring the end-effector descends exactly as the item passes through the robot's functional envelope.

Vacuum Gripper The items are manipulated by a vacuum gripper. Since I do not have a vacuum pump, I've looked for another solution to this task:

There are two pneumatic cylinders mounted next to each other. The first cylinder is connected to a valve that either lets pressure in or pressure out. The second cylinder is connected to a vacuum cup. When pressure is applied to the first cylinder it extends. As both pistons are mechanically connected to each other, the movement of the first cylinder also forces the second cylinder to extend. However, this movement generates a vacuum inside the second cylinder. This vacuum is used for picking the items.

Technical Deep-Dive: Innovations in Vacuum-Pneumatic Forensics

• Passive Vacuum Generation: In the absence of an industrial vacuum pump, this project employs a Dual-Cylinder Differential Logic. One pneumatic cylinder acts as a master actuator, mechanically driving a second, sealed "Vacuum Cylinder." When the master extends, it forces the second cylinder's piston to retract, creating a localized vacuum via volume expansion. This negative pressure is directed to the suction cup, enabling reliable item acquisition.

Engineering & Implementation

• Multi-Axis Motor Coordination:
  • Mega 2560 Computational Forensics: The Arduino Mega's high GPIO count and memory allow it to drive three stepper motors simultaneously while polling the color sensor and laser-gate interrupts. Each stepper pulse is synchronized using a high-precision timer interrupt to maintain axial alignment.
  • Structural Dampening: Precision is highly dependent on the rigidity of the Fischertechnik frame. Field diagnostics indicate that the high-acceleration "Flicks" of a Delta robot can induce harmonic vibrations; thus, the base-plate is structurally reinforced to ensure the repeatability of the 0.5mm sorting target.

Conclusion

Delta-Sort represents a landmark in Industrial Construction Engineering. By mastering Inverse Kinematics Forensics and Mechatronic Synchronization, this project proves that high-speed parallel robotics can be implemented using consumer-grade construction kits, provided the underlying mathematics and logic-flow are rigorously engineered.

---

Parallel Precision: Mastering delta robotics through inverse kinematics forensics.