Aller au contenu


TNERA

Inscrit(e) (le) 01 nov. 2021
Déconnecté Dernière activité févr. 17 2024 10:34
-----

#120282 Communication MODBUS RTU

Posté par TNERA - 08 février 2024 - 03:00

Hi, I barely have enough experience, that I doubt I could be dangerous...  :). Most of my experience was telling engineers to get it done. :D

 

Check your connection methods.  I do believe Modbus RTU is a serial protocol, and would not support any IP addresses for communications. Typically it would be RS485 or RS232.

Don't confuse it with ModBus TCP which would be caried over an addressable network.

 

Hostely, to debug, I would post your question to a good LLM like ChatGPT 4.0.  This might help you down the path to discovery.

 

Good Luck!




#120263 Finding Value in Agriculture Robots

Posté par TNERA - 04 février 2024 - 04:58

As a follow up to this thread with recent news.

 

The Agtech startup Small Robot Company failed to raise new funding and was forced to shutdown. The company is in the UK and has previously were able to raise 13m in 6 rounds. They had 50 staff. Perhaps this was some indication of how hard it is to raise funds currently due to high interest rates.  or just that the market is difficult, as stated in this article.

 

From The Robot Report:

Agtech is a competitive market

Several global startups are competing in this market for smaller format agtech mobile robots. These autonomous mobile robots are designed for smaller farms, and especially for high-value crops. These companies include Burrofarm-ngNaio Technologies, and SwarmFarm. farm-ng recently closed a $10M series A, and Burro recently announced the launch of its newest model, the Burro Grande.

This news comes on the heels of other recent bad news for some robot companies in January, including the RoboTire bankruptcy, the termination of Amazon’s acquisition of iRobot, and layoffs at Locus Robotics, Vecna Robotics and other companies.

 

Here is the full article: https://www.therobot...-shutting-down/




#120261 Des exemples de robots quadrupèdes

Posté par TNERA - 03 février 2024 - 06:12

Here is a quadruped!

Look at how fast this little guy gets around, with out bumping into anything.

 




#120229 ELROB 2024 - European Land Robot Trial

Posté par TNERA - 28 janvier 2024 - 05:32

Hi All,

 

Has anyone heard of the robot challenge ELROB?

ELROB 2024 - European Land Robot Trial

 

I guess this is similar to the DARPA robotic challenges. It is open to 'non-profits' and is for demonstration purposes.

It will be held in Trier, Germany on 24-28 June

 

unfortunately, I don't have my team together. :)

The sign up deadline is 31 January.

 

Here are the scenarios for 2024.  On the page, they have PDFs that describe in detail

 

 

 




#120217 Finding Value in Agriculture Robots

Posté par TNERA - 25 janvier 2024 - 04:49

Yes, that is a daunting task!  The amount of money required is tremendous, for larger Ag hardware startup. It really indicates that currently startups will be research projects that get spun out and funded using the existing relationships between universities and state.  Or it would be an existing industrial manufacturer that has its own R&D.  I am afraid for an individual, it would be difficult to spin up a company without already have raised Seed, or some other existing finance.

 

This is a big difference from software - since the marginal cost of developing it is mostly people-time, not Steal and Silicon.

 

I am still inspired. But, it might put Ag-Tech out of my reach. 

 

I have developed a few additional business cases.  As I said, there are a lot out there.  Ideas are cheap! its the implementation that is tough.

and Oracid said I can change my mind in a second when i want! :D:Koshechka_08:




#120205 Finding Value in Agriculture Robots

Posté par TNERA - 23 janvier 2024 - 10:13

Wow! thanks Mike, Sandro, for the excellent -sage- advice!

 

"Do you already have a clear understanding of your application?"

"try to find someone with at least some experience in your field of application"

"For you, this means at least finding a farmer willing to let you test in one of each fields"

 

I know how essential these points are. I think this is going to be hard for me. I don't live near vineyards now, and to be really successful, you need to have the robot in the mud trudging up and down the rows. Trying to SLAM in irregular settings, on hills, in different amounts of leaf canopies. And only with this experience do you learn the true understanding of the application.  This is a point of dedication. :)

 

"The goal is to swiftly enter the market and establish contact with real clients."

"You can have a nice idea, if you have no investors/clients, your won't go far.."

 

These points are extremely strong too! - and scare me. ;). All of my theoretical work has been rooted in who is the customer, and what value would they find. Being able to even get in-front of clients/investors with a good business plan all part of the work. I can work out the details, how big of a vineyard, amount of harvest, or other values streams. Until I get in-front of a vineyard manager with a working prototype... its hard.

 

I have some experience with software startups.  Writing a software MVP seems trivial compared to a Robot MVP.  What is your experience here? you had good advice on working out the application, developing the value proposition and MVP...  or is that prototype?  the adage goes Hardware start-ups are "hard"!  They require more capital, and even the costs of prototypes require some amount of seed money if it is going to be on scale.

 

and "=> Consider starting with existing robotics bases." this interesting to me too. It makes a lot of sense. And if my MVP is more about a sensor, computer vision, SLAM, or other, then using an existing base makes a ton of sense.  I just can't afford it... ha ha! (without some seed money)

 

I hope I don't come across as 'whinny'. I think there is a lot of great opportunity out there, I haven't been able to find the 'right stuff' to kick it off.   I really appreciate the conversation,   :thank_you: It's not the typical PID tuning conversation! ;)




#120203 Utilisation de Chat GPT pour les SPAMS

Posté par TNERA - 23 janvier 2024 - 09:20

ça par contre c'est assez effrayant ... Car normalement Chat GPT n'a pas de ressource pour parler de Vigibot... Cela montre un ciblage assez précis ...

C'est comme si on avait utilisé mes réponses à moi comme source pour "entraîner " ce modèle ... 
Et non je ne suis pas derrière ce message de spam ...

 

I thought this was odd, too.  I couldn't find the original introduction thread, but I thought the person might have mentioned Vigibot.  I am guessing here that the spam-puppet-master might have fed the thread into the LLM and generated the output. That might explain how it 'choose' to reference Vigibot in a paragraph.

 

I also use LLMs on a daily basis. I typically use it as a resource tool to help find details on topics that I don't fully understand. BUT you have to validate the results, everytime.  I find that ChatGPT 4.0 is pretty reliable, but will error on the side of caution if it doesn't think it can provide the answer.  Bard, googles 'older' LLM, is good to get recent event, it has internet access. But, I find that it is DRUNK a lot and provides me with incorrect details when trying to serve up data. (e.g. Asking Bard to make a list of EU companies that provide small inspection robots, it served up a list of 10 where only 3 were close to being correct and the others were not EU companies, or did not supply robots at all. ).  therefor, I think you really have to use them like a tool, and try not to exceed the capabilities.

 

What I worry about, is when the SCAMMERS start using LLMs to write better Scam emails.   :nono: 

hopefully, i will have a dedicated AI Agent that can spot these first!




#120188 Utilisation de Chat GPT pour les SPAMS

Posté par TNERA - 21 janvier 2024 - 10:25

"The observation is fascinating. It parallels the concept of the 'Uncanny Valley,' where we, as humans, possess the ability to discern simulants in our midst solely through their textual expressions."  (ChatGPT modified my text(
- ha ha, thanks ChatGPT!
 
I do really find it interesting, that we CAN sense out that someones text is 'generated'.  For now at least, that responses from LLMs are too 'correct' and ultimately fail the Turing Test.



#120159 Finding Value in Agriculture Robots

Posté par TNERA - 15 janvier 2024 - 11:53

Hi all, (sorry in English, but most browsers can translate really well)
 
I wanted to start a discussion about robot businesses. I feel like we are entering the next age of robotics. A lot of the components necessary for a success are becoming present and easy to obtain such as Open Source technologies like ROS, high-power low cost computing, cloud infrastructure for data, low cost sensors, lower cost batteries, good communications. This would allow entrepreneurs to build robots to go after fantastic emerging use cases! Right?
 
My dialog prompt is:
  • Will the robots be more successful by automating redundant/dangrous/boring work?
  • Or would they really require strong value added services like data intelligence, or the 'killer app' syndrome?
 
As an example company, I present Burro ( burro.ai ). Burro is a US company providing a robotic platform that works in high value row crops such as vineyards for grapes, blueberries, and plant nurseries. The company has been around for 4-5 years, and has recently raised about 25 million in Series B funding (these numbers are just estimates from memory). Burro operates in the US market, and perhaps South America. They have not entered the European market (so they are ripe for competitors here!!)
 
gen8_front_side_withUI_fixed-copy_smaller.png

 

Here are some of their training videos
 
 
Row Navigation - interesting User Interface, but basic capability
 
It seems that they have a good use case for transporting picked fruit from the pickers in the vineyard to the packing tables at the ends of the rows. It seems this would save time, and reduce the number of laborers required.
 
But here is a video (with very bad sound), that demonstrates the "scouting" use case. Here using software from a Computer Vision company, they are counting clusters of fruit. This is typically done to provide better estimates for harvest. This allows the vineyard manager to make proper arrangements for crates, workers, number of trucks, estimates on weight etc. This is a value add use case, and it looks like it would be an additional cost potentially.
 
"Scouting" with Bitwise (Computer Vision)
 
The Burro has the list price at the same cost as a Sub-Compact Tractor. ~15,000 - 25,000 USD. It also comes with a monthly "automation" fee of 200-300 USD. Over a four year period this could have a total value of about 100,000 USD. This probably does not include the software from the Computer Vision software. Which leads to may question.
 
What do you think?
  • Will it be the labor replacement?
  • or the making Ag more digital? Meaning that the value is best found in areas like Computer Vision, and sensors, and data collection.
  • Or Are we still missing the 'killer app' that will allow this robotic platform to take off?

Image(s) jointe(s)

  • gen8_front_side_withUI_fixed-copy_smaller.png



#120122 Au bistrot du coin ...

Posté par TNERA - 07 janvier 2024 - 02:54

Bonne Année!   Happy 2024, All!

 

The Robot Report has an article on 5 Trends in Robotics for 2024.

 

5 robotics trends to expect in 2024

 

I would agree with the projections, especially in seeing an expansion in Mobile Robot Use Cases. I am hoping we start to see more robots in every day use.

 

One question though, at the end of the article, they suggest that Rust will become the language of choice (or at least more wide spread) in robot software.  What do you think? Is anyone using Rust, or Rust with ROS? 




#119647 Des exemples de robots quadrupèdes

Posté par TNERA - 16 octobre 2023 - 04:30

Quel sera le robot de l'avenir ? Quadrupède, bipède, roulant ? 

Et si c'était les trois en même temps ?

 

yes! I think the robot labs at ETH already have that answer as well.

 

 

Now if we can put x6 7-dof arms on it we will have something really cool/scary.  :D




#119635 8DOF - Q6 - Un quadrupède de taille moyenne

Posté par TNERA - 15 octobre 2023 - 12:52

 

Je vais profiter de ta période de retrait du quadrupède pour tenter de descendre en dessous des 9s/10m et venir à la TRR avec beaucoup d'ambitions...   :laugh1:

 

ha ha, Me too!  :D




#119497 Au bistrot du coin ...

Posté par TNERA - 25 septembre 2023 - 04:48

The improvements are impressive! The focus on better actuators is starting to pay off. 

 

I am not personally a big fan of humanoid robots. I understand the benefits of being able to work in a 'human' world. However, I find the interest on it taking away from other applications for robotics. There is so much more we could do if it were not specifical the size and shape of a human.




#119430 Au bistrot du coin ...

Posté par TNERA - 17 septembre 2023 - 10:06

This interesting is, it reminds me of the concept of a "smart wheel" or what I would call an "active wheel". The concept has been around, to have 'feet' on the wheels that adjust to the changes in the ground. I know James Bruton made a 'Pedrail Wheel' (invented in 1903) type devices but it is very clunky. 
 
Pedrail_wheel.jpg

 

However, my inspiration comes from the description in 'Snow Crash' by Neal Stephenson:
 
Smartwheels use sonar, laser range finding and millimeter wave radar to identify mufflers and other debris. Each one consists of a hub with many tiny spokes. Each spoke telescopes into five sections. On the end is a squat foot, rubber tread on the bottom, swiveling on a ball joint. As the wheel rolls, the feet plant themselves one at a time, almost glomming into one continuous tire. If you surf over a bump, the spokes contract to roll over it. If you surf over a pothole, the rubber prongs probe its asphalt depths. Either way, the shock is thereby absorbed, no thuds, smacks, vibrations, or clunks will make their way into the plank or the Converse hightops with which you tread it.The ad was right - you cannot be a professional road surfer without smartwheels.
 
And another:
 
“The smartwheels of her skateboard, many, many spokes extending and retracting to fit the shape of the ground, take her across the lawn like a pat of butter skidding across hot Teflon.”
 
Any other good examples of this technology applied? Or is it still before its time?
 
 
Btw, I am just back from Provence, it's a lovely place!



#119052 Des exemples de robots quadrupèdes

Posté par TNERA - 23 juillet 2023 - 12:12

This Chinese consumer robot company has introduced it's newest model.  Unitree GO2.

 

It hop's, jumps etc...  also integrating lidar and ChatGPT.