Conveni Pizza

Students: Yoeri van Polanen, Luthfiandra Ardana, Dennis Velthuijzen, Muhammad Gamal Alfauza

Lessons learned

The robotlab had several limitations that changed our simulated setup from the setup at Conveni:

  • There is one small robot arm available with limited reach and cannot reach the ground.
    • It was decided to use the crates and dolly as offset instead of a block to ensure that any movements of the dolly’s and crates would be simulated.
  • There is limited space in the robotlab.
    • Only a dolly and 10 crates were brought in for testing. The official setup has rollies (double sized rollies) and about 44 crates total.

This caused our setup to be a lot smaller than the one at Conveni.

In the program you can enter the amount of crates available for testing purposes and the program only works for dollies at the moment. In practice, with a bigger robot arm, you could set the amount of crates in each cycle to a static 22 crates for one side of the rolly and rerun the program again with an offset for the next 22 crates.

The program allows the user to set the time taken for each movement action, allowing the robot to scale it’s speed automatically to different distances caused by height difference at the pickup and dropoff points. However, this speed is limited by the maximum speed hardcoded in the robot. The current maximum speed is around 12 seconds per crate. (2 pizza’s per crate, 6 seconds per pizza. A total of 10 pizza’s per minute.)

This maximum speed also prevents us from completing the preferred speed of 24 pizza’s processed per minute. As maximum torque, acceleration and collision errors occur and certain processes, like the crate locking in place with gravity and the pizza’s falling into the crate take time.
To solve this issue, either the maximum speeds of the robot need to be much higher and optimized, or a secondary arm needs to be used.

Another problem that occurred is that the second pizza is not allowed to ‘slide over’ the first pizza when coming off the conveyor belt. The sliding action can cause the second pizza to get stuck on the first pizza, preventing it from falling into the crate properly. To solve this, we first catch the first pizza. Afterwards we move the crate under the conveyor belt to catch the second pizza.
Then we slowly move the crate backwards to make sure the second pizza is in the crate properly.

To help with the catching process and the reduced pizza’s per minute speed, a buffer was added to the conveyor belt. However, it was discovered that the robot computer (PLC) only outputs 24VDC voltages for signals, switches and other external components. This can be brought down with a single 24VDC to 5VDC converter, but a second one converter cannot be connected. The robot will give out errors then. This may be a bug in the Doosan system. A workaround was made with relays.

Lastly, the robot movement of the robot arm causes the table it is attached to to move slightly when in use. This forced us to attach the pickup point and the dropoffpoint to the table to move with the robot. In a practical scenario, a bigger robot arm just needs to be bolted to the ground and the pickup and dropoff points need to have clear slots for the rollies to slot into.

To summarize:

  • The simulated setup is smaller due to space constraints and limited robot arm reach.
  • The program has been made modular to easily handle more or less crates.
  • The program can automatically change it’s movement speeds depending on the required distance to travel.
  • The speed of the process cannot reach the preferred 24 pizza’s per minute due to limitations of the robot speed and available tools.
    • The current maximum speed is 10 pizza’s per minute.
    • Perhaps a second arm or faster robot can resolve this.
  • The pizza’s are very bad at sliding over each other, so the crate must be placed under the conveyor belt to catch the second pizza.
  • The robot PLC outputs only 24VDC. Scaling down this voltage can be done, but is hard due to a possible bug in the Doosan system.
  • The pickup and dropoff points need to be fixed and calibrated for the first setup. The robot arm must also be bolted to the ground.