Snake-like Robot Modeling and Navigation (Project 224)

[robertogl]

Contribute to the discussion by asking and/or answering questions, commenting, or sharing your ideas for solutions to project 224

---

🤖 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.

[Nachiket497]

Hello, my name is Nachiket Sawwalakhe and I am a student at VNIT Nagpur. I am interested in working and collaborating on this project. 

[SafalRaj10]

Hi, This is Safal Raj, a Mechanical Engineering student at IIT Patna. I'm intrigued and will be contributing to this project.

[grsasaki-tmw]

Hello @Nachiket497 and @SafalRaj10, we do appreciate your interest in this project! If you have any questions or points you would like to discuss, I welcome and encourage you to post here. Best Regards, Gen Sasaki

[Nachiket497]

I and @amit-420 have created a simulink model a snake and we want to run it on ground plane. How to add a ground env along with the snake model ? can you please provide some resources on this?
@grsasaki-tmw

[grsasaki-tmw]

@Nachiket497, assuming you are using Simscape Multibody as the project recommends, you can use the Infinite Plane block to define the ground plane. (https://www.mathworks.com/help/sm/ref/infiniteplane.html)

[Nachiket497]

Will check on this.

[amit-420]

@grsasaki-tmw we are trying to implement infiniteplane in snake model. As of now we have connected "R" of the block to world frame and "G" to the contact force block as mentioned in the documentation but we cannot find how to move ahead. Could you provide any example where infinte plane is used with a CAD model with multiple contact points ?

[bhavikpatel2]

Hi, I am a research scholar at IIT Guwahati and am interested in working and collaborating on this project

[bhavikpatel2]

Hi @robertogl
I am fine on Friday. I am following +5:30 GMT. I will connect with you anytime on Teams/ Google Meet/ Zoom whatever suits you.
Email id: bpatel@iitg.ac.in

[bhavikpatel2]

Hi @robertogl
It was a nice conversation with you.
Finally, I found the problem. The transformation sensor doesn't have full-angle information. It has only absolute value and therefore, the head angle controller failed. I have rectified that problem, and the robot tracks whatever reference angle is given. Finally, I implemented a head controller. You may find the final version of implemented head controller in [Robot_head_control/Snake_Robot_Head_Control.slx].

[robertogl]

Awesome! I am so glad to hear that you found and fixed the problem. It sounds like you have done with the project. Is there anything else you are planning to do before submitting the solution? Otherwise, please go ahead and complete the submission form.
Thanks!

[bhavikpatel2]

Yes, I will submit it by making readme,md file and comments within two week.
Thank you.

[bhavikpatel2]

Hi @robertogl

I have submitted the solution and updated the Readme.md file. I tried to make it as simple as possible for learners. If any upgradation is required, kindly suggest me.
Thank you for being so supportive in completing this project.

[OdlLed234]

Hi. I am a student at the University of the Witwatersrand, Johannesburg, and interested in working and collaborating on this project

[Antoine-ms]

Hi, I am Antoine Masson, a mechanical and control system engineering student at INSA Rennes, France. I am interested in working and collaboring on the project

[Antoine-ms]

I uploaded 3 versions of my anisotropic model on my repository (3_1 ; 3_2 ; 3_3). I checked the reference frames again and I tried to swap them, but I still couldn't manage to get my snake to move.

[Antoine-ms]

Hi @robertogl

I finished the design of my controller, I tested it on several scenario and everything works fine exept for one last problem :
The desired direction of the snake is computed as an angle in radian between -pi and pi, which raises the problem of the discontinuity. Do you have any idea how I can handle that ?
Thanks

[robertogl]

The error to feed back has to be calculated properly so to identify those jumps and unwrap their values when bigger than a certain threshold ('pi') by adding 2pi with the appropriate sign. For these cases our unwrap function will be useful.

[Antoine-ms]

Hi @robertogl

I am struggling to get the unwrap block to work. It seems like it has no effect on the signal. I tried to play with its parameters but I couldn't make it work. Do you think you could take a look on that ?
Thanks

[robertogl]

I am having some issues in running the model. But have you tried applying the unwrap only to the error angle?

[Alexan41]

Is the competition still on?

[robertogl]

Hi @Alexan41,

You can easily use this template. Just click on 'Use this template' and then 'Create your repository'. Once you have done that you just need to change the name of the repository and the owner (you will see your profile name in the drop down menu), and then just click on 'submit'. In the new repository you can literally drag and drop the files you need to share.

That's a good start to share your initial files. Let me know if you need anything else.

[Alexan41]

It is done. I have submitted my work.
Thank you Sir.

[robertogl]

Thanks @Alexan41! I have received the submission, but I am bit behind with reviews because of other commitments. I will get back to you as soon as possible. The video looks great btw 🙂!

[Alexan41]

Thank you sir!

[Alexan41]

Hi @robertogl
I think you forgot me, I'm still waiting for feedback on my project.

[RKommX]

Hi, I'm Rajdeep Paul, 2nd year aerospace engineering student at IIT Bombay. I wish to spend my semester break working on this project.

[RKommX]

Hey @robertogl

I figured the spatial contact forces may have been creating problems, so I removed them and used planar joints instead of 6-DOF joints. I also used a LOS control for the robot. The simulation runs to completion, but the robot shoots away, so I think there is some unstability in the system.

I've updated my repo. Please take a look if you have time and let me know what can be done.

1-s2.0-S1367578817301050-main (1).pdf
This is the source I'm referring to.
Thanks.

[robertogl]

Hi @RKommX,

I have tried to run your model but I got several errors. I fixed a coupled but not all apparently. Can you check if your model works fine after you download it from the repo in a fresh folder?

Also, I suggest you to take a look at the Point Cloud block or the Sphere block to model the geometry of each link of the snake that makes contact with the ground. Each link would have a point cloud associated with it, and then that point cloud has a contact force block between it and the ground. Then tune the parameters appropriately. You might be able to get away using a sphere instead of a point cloud. Here is a decent example.

Another option is taking a look at Bhavik's submission. Perhaps you can build your solution on top of that and create a more complete simulation of an autonomous snake robot that includes sensors and autonomous navigation in a more complex environment. Up to you.

[RKommX]

Hi @robertogl,

I finally got it to work. I changed the gain value from the LOS block and a few other parameters. It moves smoothly, but the only issue is that it seems to move slightly downwards (very slight compared to its forward motion). I have updated my repo, so please check if it's alright.

I do plan on making an underwater version. I will also try a sphere link. For now, I want to brush this up and try to reduce its deviation as much as possible.

Thank you.

Edit: Reattaching the link, previous one gave 404
https://github.com/RKommX/Snake_Robot

[robertogl]

Hi @RKommX, sorry for the delay. Please allow me a few more days and get back to you.

[RKommX]

No issues. I've already been bogged down by academics, so I haven't been able to work on it for a while.