Adaptive Palletizing with Simulation Optimization (Project 254)

[robertogl]

Contribute to the discussion by asking and/or answering questions, commenting, or sharing your ideas for solutions to project 254

---

🤖 An AI bot may reply to your comments with suggestions and guidance. It will not solve the project for you. Your comments may be processed by a third-party LLM. To direct a question to a specific person, tag them with @username.

[Tanmaey45]

Hii! I am Tanmay. A mechanical engineering student at the Indian Institute of Technology, Bombay. I am interested in working and collaborating on this project.

[Tanmaey45]

I am stuck in the robot palletizing example. In the helpercontopptraj block, the source code uses 'contopptraj,' a standard optimization function which takes 4 parameters or values as inputs. Running the simulation, it asks for solver name to be specified, but we can't since contapptraj takes only 4 values. So its a loop of error, if we do one other becomes error.

[robertogl]

Hi @Tanmaey45,

Do you have the Optimization Toolbox installed? Please make sure you have it installed.
If you do already, please paste the error that you get and the snippet of the code the error refers to.

[Tanmaey45]

Got it! Thanks

[aastha51551]

The sign up link for this project isn't working. Can you kindly look into it? I have added an issue as well.

[robertogl]

Yes, we are experiencing some network issues. Please use this form as an alternative option to sign up. Just add project title and number (254).

[Tanmaey45]

Hello, The issue got solved. I just wanted to ask whether the challenge wants us to solve optimization for 2D arrangement on the palette for variable box sizes or stacking layers also, i.e. solving and simulating a 3D packing problem Regards, Tanmay

On 2025-05-16 20:30, robertogl wrote: Hi @Tanmaey45 [1], Do you have the Optimization Toolbox installed? Please make sure you have it installed. If you do already, please paste the error that you get and the snippet of the code the error refers to. -- Reply to this email directly, view it on GitHub [2], or unsubscribe [3]. You are receiving this because you were mentioned.Message ID: ***@***.***>

Links: ------ [1] https://github.com/Tanmaey45 [2] #127 (reply in thread) [3] https://github.com/notifications/unsubscribe-auth/BFFMEJDDUTTT2XJVEU5IUWD26X4RPAVCNFSM6AAAAAB2ZQBDZKVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMJXGIZDMNY

[robertogl]

It should be 3D, but if too difficult 2D is a good start.

[Tanmaey45]

I am getting an error while grasping boxes from pickup locations other than the spawn location; the boxes are getting gripped higher and not at the end effector.

[robertogl]

Does the pickup location correspond to the top part of the box or is perhaps at the center or bottom of it?

[Tanmaey45]

It is even above the top. I have kept an offset so it doesn't collide while in trajectory

[robertogl]

It sounds like a transformation error, but it is hard to give more details without looking at your model.

[Tanmaey45]

How to add already placed boxes as collision objects in collision environment

[MateoF818]

Hello, I've been working on the optimization problem for the boxes and I haven't been able to find literature on any sort of palletizing "good practices" for the handling of differently sized boxes. My question is, are there any specific constraints put on how smallers boxes are arranged vs larger ones in the context of heterogeneous palletizing? Thanks!

[robertogl]

Hi @MateoF818,

Yes, real-world palletizing (especially with mixed box sizes) usually imposes stability and handling constraints. Here some guidelines I found online:

• Weight & center of gravity
  Heavier/larger cartons go low and centered to avoid tipping.
  Robopac USA – Load Stability Guide

• No overhang
  Boxes must stay within the pallet footprint; overhang weakens edge support.
  OSHA Training – Stacking Materials Safely

• Minimum vertical support
  Each upper box should have ≥70% of its base supported (or at least 3 corners).
  ScienceDirect – Mixed-case palletization problem

• Stacking patterns (column vs interlock)
  Column = stronger vertically; Interlock = better lateral stability.
  Boxes & More – Pallet Stacking Patterns

• Layering rules for mixed sizes
  Group small boxes into “super-boxes” and keep layers even; add tie-layers.
  SpringerLink – Mixed-case palletization
  UWSpace – Palletization thesis

• Height & handling limits
  Caps on stack height; blocking and interlocking per OSHA.
  iGPS – Pallet Safety Guidelines

Even the paper we suggest in the project description page follows those palletizing guidelines.

[MateoF818]

Thanks!

[zappateinstein]

Hi everyone, i am currently a student at the university of technology of Compiegne in France and i am interested in working and collaborating on this project

[github-actions[bot]]

Thanks for your interest. A good next step is to open the project page and read the description, then register your intent using the sign-up form linked there so you can receive the project details. If you want to collaborate with others, you can mention the MATLAB/Simulink areas you’re comfortable with (for example Robotics System Toolbox, Simscape Multibody, or Optimization Toolbox) and the kind of contribution you’d like to make. For general preparation, the MathWorks repository of student resources and the project’s repository guidelines are good references for setting up a clean, runnable solution.

---

🤖 This is an automated response. For further assistance, please tag a project mentor or visit the wiki.

[zappateinstein]

I have used a bit Simscape Multibody but i want to know about the toolbox. As i want to work and collaborate to the project, the access of the toolbox is given by Mathworks, or i have to subscribe to it myself?

[github-actions[bot]]

Toolbox access is not provided automatically by the challenge program. You’ll typically need a MATLAB license that includes the required products through your university, personal license, or trial/student options, and the project page will list the toolboxes/add-ons you should verify before starting. For this project, check whether you have Simscape Multibody and any related tools such as Optimization Toolbox or Robotics System Toolbox if the description mentions them. You can confirm installed products with ver in MATLAB, and more info is available in the MathWorks product pages and documentation at mathworks.com/help.

---

🤖 This is an automated response. For further assistance, please tag a project mentor or visit the wiki.

[omkarputti]

Hii! I am Omkar. A mechanical engineering student at the Gogte Institute of Technology,Belagavi. I am interested in working and collaborating on this project.

[zappateinstein]

Hi, everyone, i began on the problem with the implementation of an excel files that will input the characteristics of my boxes, by using the initFcn program in the simulink model given of the callbacks of the model properties.



And i create this block for displaying the boxDim and boxPosition in the simulink trajectory planner here



However, i encounter many problems and errors even though i change the size with signal specification block or in the function used, please can with help me with this problem.
Here is the errors i have when i launch the simulink model.
Simulink cannot propagate the fixed-size mode from the '[Output Port 1](matlab:Simulink.internal.open_and_hilite_port_hyperlink('hilite', ['PalletizeBoxesUsingCobot/Trajectory Planner/MATLAB Function'], 'Outport', 1);)' of 'PalletizeBoxesUsingCobot/Trajectory Planner/MATLAB Function' to the '[Input Port 1](matlab:Simulink.internal.open_and_hilite_port_hyperlink('hilite', ['PalletizeBoxesUsingCobot/Trajectory Planner/Signal Specification4'], 'Inport', 1);)' of 'PalletizeBoxesUsingCobot/Trajectory Planner/Signal Specification4'. This input port expects a variable-size mode. The fixed-size mode originates from 'PalletizeBoxesUsingCobot/Trajectory Planner/MATLAB Function'. Either replace 'PalletizeBoxesUsingCobot/Trajectory Planner/MATLAB Function' with a block that outputs a variable-size signal or replace 'PalletizeBoxesUsingCobot/Trajectory Planner/Signal Specification4' with a block that expects a fixed-size signal.

[github-actions[bot]]

Your issue is likely a dimension/mode mismatch: the MATLAB Function block is producing a fixed-size array, while Signal Specification4 is configured to expect a variable-size signal. A good first step is to decide which one you actually need—if the number of boxes is known from the Excel file, it is often simpler to keep boxDim and boxPosition fixed-size and set the Signal Specification dimensions accordingly; if the number of boxes changes during simulation, then the MATLAB Function must explicitly output a variable-size signal using supported variable-size syntax. In the MATLAB Function block, check the Edit Data / Symbols pane and make sure the output is marked with the correct size and VariableSize setting, and confirm the downstream block matches that setting. Also verify the data loaded in InitFcn is a numeric matrix/array with consistent dimensions; for debugging, use size(...), whos, and a simple disp(...) before connecting it to the model. Documentation to review: Variable-Size Signals in MATLAB Function Blocks and Signal Specification block in MathWorks Help.

---

🤖 This is an automated response. For further assistance, please tag a project mentor or visit the wiki.

[zappateinstein]

i continue to have the same problem, i tried anything. I try to include the From Spreadsheet Block but i continue to have problem from Self collision. In fact i my initFcn i create the variable of boxDim and boxPosition in order to create the first box.


And i used from spreadsheet to obtain the value of boxDim and boxPosition at each trigger but i got a problem.



That the variable when i try to get the values from a To workspace block or a display block continue to be [0 0 0] and not [-0.65 0.65 0.6] for the example that i am trying. Then it causes problem of self collision with the robot.


What i have to do here to remove this problem?

[github-actions[bot]]

The [0 0 0] value usually means the downstream block is being evaluated before the spreadsheet data is available, or the signal is not being updated in the way you expect. Try verifying the data loading order first: InitFcn should create boxDim/boxPosition in the base workspace before simulation starts, and the spreadsheet block should output a numeric array with a fixed, known size that matches the block ports. For debugging, place a Display or Scope immediately after the spreadsheet block and confirm whether the signal changes there; if it does, the issue is likely in how the trajectory planner or collision block reads it, not in the spreadsheet itself. Also check the self-collision setup in the robot model: if the box position is treated as an obstacle at the robot origin because the input remains zero, the collision checker will trigger even though your intended data are valid. The most useful docs here are From Spreadsheet and MATLAB Function block data typing / signal sizing in MathWorks Help.

---

🤖 This is an automated response. For further assistance, please tag a project mentor or visit the wiki.

[zappateinstein]

that is the problem, when i launch the simulation, i don't see anything from the display, nothing. What can be the problem?

[github-actions[bot]]

If the Display stays blank, the signal likely isn’t being produced at all rather than being “wrong.” First check that the From Spreadsheet block is actually executing: confirm its Sample time, file path, sheet/range, and that the model is not stopping earlier because of an initialization or dimension error in another block. It also helps to test the spreadsheet block in isolation by connecting it directly to a Scope or Display, and verify the loaded variable with readmatrix or readtable in MATLAB before the model starts. If the block is inside a triggered/conditioned subsystem, make sure the trigger really fires; otherwise no output will appear.

---

🤖 This is an automated response. For further assistance, please tag a project mentor or visit the wiki.