11 réponses à ce sujet

[Mike118]

(French version here)

 

Hello everyone,

 

This year, I am participating again in the Coupe de France de Robotique, but this time, I'm starting with a new team composed of

pat92fr hdumcke  (whom I both met during the TRR), and myself.

Pat chose the name Mars Attacks! in reference to the movie of the same name because the theme for the 2024 edition of the Coupe de France de Robotique is "Farming Mars," revealed in September 2023.

 

It's the first participation for pat92fr and hdumcke (yes, I roped them into this, but I promise they are willing participants and not being held hostage ). Since it's their first time, we are starting a bit from scratch, but since we are not beginners in robotics, we're progressing quite well. As they are both in the Paris area and I'm in the south, near Biarritz, most of our sessions are conducted remotely, which can be a challenge, but we're managing quite well.

 

To provide a brief update on where we stand at the moment:

 

Mechanically:

 

We are working on a mecanum base with a lidar positioned quite low to hit the table edges for localization (based on my experience with vigibot). Counter-axes are added at the wheel level in the protective fairing to relieve the motor reducers a bit. The front motor train is mounted on a pivot linkage to keep all four wheels in perfect contact with the ground.

 

Some photos:

 

Assembled base from the top:

Assembled base from the front:

Highlight of the front train pivot without wheels:

Top view of the motor support integrating bearings and counter-axis in the wheel fairing:

Front view of the motor support integrating bearings and counter-axis in the wheel fairing:

 

For this mecanum base, we are using:

• 4 DJI M2006 motors with their C620 controller
• A set of 4 mecanum wheels
• 8 flanged bearings (8mm)

 

Several 3D-printed parts (PETG), (the files will be shared) and plenty of screws and fasteners 

 

Software-wise:

 

We have implemented a visualization tool "from above" that allows, among other things, controlling the robot and logging data. This visualization tool, developed in Java with Processing, can connect directly to the robot during tests or to other simulation tools that have also been developed.

 

Here's what the tool looks like at the moment:

 

This tool will be shared with various examples on the visualization tool's Git :  (Basic examples have already been shared, with the idea of providing building blocks so that everyone can develop the tool that suits them from examples) 

Robot Code:

 

Regarding the robot code, for now, we have mainly focused on robot control and localization on the table. The system runs on an STM32, and data is sent to the visualizer via WiFi using an ESP32. Currently, we have a working system, although not perfect yet, connected to our visualizer. During tests, we get results like in this video: 

In this video, the robot is moving on a table, and what we see is the data sent by the robot. It displays where it thinks it is on the table, and in red, we see the lidar points. 

Around 1 minute and 50 seconds, the robot touches an obstacle and loses its localization. We need to work on detecting opponents, and it's in progress, but the results are already quite interesting. 


We plan to publish a lot more, but we need to make progress first, so stay tuned for more soon!

[Mike118]

Hello everyone!

 

We would like to share an update regarding our participation in the French Robotics Cup.

In terms of progress, the end of this year has been very intense, and although we haven't been able to make as much progress as we would have liked, we still have a few points to show you!

 

Firstly, we continue to refine our mecanum base.

Here is how the current version of the base is assembled and wired:

 

 

 

 

 

 

 

While we work on the final version of the robot base, we are also improving the localization algorithm and conducting movement tests with the prototype base.

 

Here are some video footages:

 

Video showing the robot  on the table:

 

Robot showing what display the Visualizer tool:

 

And finally, today is the day we need to submit our scientific project report and communication plan! As "veteran participants" (over 30 years old), we are registered in the "Legend" category and have a few additional tasks, including promoting the event

 

Here's a preview of our report:

 Mars Attacks! Scientific Project and Communication Plan.pdf   2,35 Mo   66 téléchargement(s)

 

We hope that among our readers, we can inspire a few people to participate in the Cup next year! =)

We'll be back with more updates soon!
Meanwhile, we wish you all a fantastic 2024!

[TNERA]

Wow!  That looks exciting and challenging!  Good luck guys on your development, I am looking forward to seeing the progress.

 

You all really have a lot to integrate together to perform the tasks and meet the design constraints. After reading the rules, I see you have to add a grippers, beacons, emergency stops, etc. Lots to add on. Do you plan to use the 'local computing space' to provide more visibility? will there be more than one robot?

 

The base looks slick, and performs very well in the video. I had to do some quick research to learn what "holonomic" means!   Are you driving all of the SLAM from the lidar? to include location and the orientation of the base?

 

this will be very interesting!

[TNERA]

It has been really quiet... 

how is the progress on the robots?

[Mike118]

Hello everyone !

A little late here is a little update for our progress this month of January:

We made some pcb:

=> A board based on teensy 4.1 integrating quite a few UART sockets, a CAN socket, and electronics to control feetech motors.

=> A card for the robot's power supply integrating enough to connect an On Off button, an emergency stop button, the battery and an external power supply, as well as a current and voltage sensor.

=> The PAMI card which allows you to control 4 DC motors with encoder and which integrates, among other things, an IMU. ( You can also see the current state of the Pami prototype)

We have also made progress on the path planning of the robots with the implementation of an A* algorithm on an occupancy grid and we display the result on our visualisation tool. 

 

[Mike118]

Still running late, but making progress!
Yes, the Belgian Cup is this weekend.

Here's a sneak peek of the first version of our technical poster:

The bulk of the main robot's mechanics is okay, although there are still a few points to review/improve
Attached are a few more images:

 

 

At the back, there's the screen displaying the visualization tool and allowing selection of color and starting position =)

 

On the Pami side, we're a bit late:

 

We should have only 2 ready for this weekend!

But it will be good preparation for the French Cup in 3 weeks!
 

Video : 

[TNERA]

That is one fine looking robot!  

 

how did it go this weekend in Belgium? 

[Mike118]

Result of last weekend's Belgian Cup:
this training was really useful. We identified some mechanical issues that were not optimal and that we need to correct.
We validated that our robot is capable of positioning itself correctly in the right place on a table, but we still need to fine-tune our trajectories a bit, and we need to develop our strategy, especially against aggressive robots that will be starting on our side =)

But above all, we need to have more Pami!
We only had one, and that is clearly not enough to compete with the number 1 teams who had 6!

We still finished 4th out of 26 teams behind the Belgian number 1 Mons'tellair and two other French teams who trained in Belgium like us: TDS and Les Karibous. We had a rather good final score despite all the little problems we encountered during the matches.
Here are the videos of the finals


 

[TNERA]

Wow, I am impressed.

 

points are points, but you guys were potting the plants in the buckets, while others were throwing the plants around!

 

Yes, a few points here and there on strategy and you all are going to be great contenders. 

 

Thanks for sharing the link, its really interesting to see the different strategies. 

 

All the best on your run up to the French Cup!

[TNERA]

Hi Guys!

 

Any news on how the Cup went for you and your other sponsored competitors?

 

I picked up a little from the other post that it was a good first run?

[pat92fr]

Hello,

 

When we registered for the CDR in Legend, we submitted a scientific project with the goal of publishing a method for simple and low-cost absolute localization.

 

After the Coupe d'Ile de France, we finalized our paper. In it, you will find a description of our localization algorithm using a truly affordable LIDAR. This technical subject kept us busy throughout the 2024 robotics season and posed quite a challenge for us.

 

From the first matches in the Coupe de Belgique, where the algorithm was not yet perfected but miraculously worked, to the matches we almost entirely lost due to localization issues in the Coupe de France, we continuously improved our method. Finally, our localization proved to be reliable and efficient in Brittany (Breizh Ty Geek) and in Ile de France, just today!

 

We hope this technical contribution will be useful to the community! Please feel free to comment on our paper.

 

 Enhancing_Robot_Navigation_with_a_Low_Cost_LIDAR_and_Absolute_Localization_Algorithm (1).pdf   1,69 Mo   91 téléchargement(s)

[TNERA]

Kudos! to you.  Thanks for putting the work and research in and making it for the community!

congrats to the team!