Newbie here looking for some advice.
I have just started a robotics project that involves creating a robot that can autonomously move to collect 3 balls (tennis ball, squash ball and table tennis ball - all different colours), collect these balls, then navigate over a seesaw which is yellow in colour to a ball deposit zone. The robot must then drop off the balls in this deposit zone and move to a system end zone on the same side of the seesaw (only requiring crossing the seesaw once). I have attached a little CAD model of what the playing field looks like.
I have stumbled across this device called a PixyCam which I believe CoreElectronics sells. What seemed really intriguing to me was that it was able to find and move toward different objects and colours autonomously without requiring too much in depth computer vision coding (which is a big plus for me).
I am still not entirely sure how it works and need a bit of guidance as to whether something along these lines would be viable for the project that I am working on? Would it be able to detect these balls, move to their position and then execute some code that will activate a mechanical system to collect these balls, then find the seesaw, cross the seesaw and search for the ball deposit zone. After which it executes another line of code to deposit the balls?
That is a pretty ambitious project for a “newbie”. I wish you well.
Regarding the "PixyCam. This looked like an interesting device so I had a look.
It would seem that the item sold by Core may be “PiCam” which looks like just a platform to mount a “standard” (is anything RPi “standard”) RPi camera.
The PixyCam looks like it is sold by Tribotix who are almost next door to Core in the same street. Coincidence???. Looks like a clever little beastie and seems quite reasonably priced considering all the smarts involved.
Cheers Bob
Hi Alex
I think Core’s search system needs a look or they call the product by the same name as everyone else.
I did a search on what you (and Tribotix) called it, ie; “PixiCam” and it insisting on changing it to PiCam so without knowing exactly what Core have chosen to call it I assumed they don’t stock it.
How many stock items did you search to find it??
That scenario is as bad as searching for an item type and finding about 4 or 5 but another few buried in hundreds more which bear no relevance to the search criteria. Something like IKEA where you have to walk past everything in the store to find the exit.
Simply Googled “PixiCam” and they came up. Core didn’t.
Cheers Bob
There is some examples on the Pixy Wiki that demonstrate line tracing, with some modification it could be adapted to your use case (don’t be afraid to use things like ChatGPT for help with coding and to give you some ideas rather than trawling websites for crumbs for hours on end)
Yeah it gets a bit like that doesn’t it.
I think I just got a little bit lucky and stumbled across it, probably from a similar thread on this support page.
The link you shared looks very interesting! That may even allow me to completely avoid the seesaw completely and just trace the right edge up to the balls, collect the balls, then trace the top edge and navigate around the seesaw.
Have you had any experience working with the Pixy Cam at all? What I just want to get clearer is how easy it is too adjust the tracking systems for my usage. All of the information that I have found on it seems to be that it can locate and follow a specific moving object.
In my case however, I need it to locate the balls (which are stationary and in a known location relative to the start) and then execute a code block that deals with some additional hardware to collect those balls, then continue about tracking other objects around it to lead me to the ball deposit zone (whether that is over or around the seesaw).
What I don’t want happening though is to not be able to modify the code within the Pixy Cam and just have it follow the non-moving balls around or not be able to execute other code as well?
Sorry if I haven’t really made my question too clear but that’s really what I’m trying to understand at the moment.
Looking at some of the documentation on the Pixy Wiki it would appear that it is easy enough to teach it new objects, the only issue that you may run into is if the squash ball is black as it won’t have a distinct hue.
