Damn! - the good news is that I got a bit of free time over Xmas to work on my robot - even had 1/2 hrs usefulish practice against it [still very much cobbled together! ]- unfortunately the bad news is that one, or possibly two of the circuit boards that were driving the servos and motors are now knackered! - i think they possibly overheated, or maybe the current overloaded them but now they are pretty much dead! - its not a show stopper though, going to order some more in the new year [a hefty £40ish]- and things should be back to where I am [just got to figure out what happened so i dont burn the new boards!]

BS.

Ok - had a little chance to work on the bot over the weekend. Still no videos/images sadly, but they are on the way.

I Was working on the feeding mechanism over the weekend, the main problem is that the head will be twising and turning over 3axis, whatever feeding mechanism used will have to get a ball from the ball hopper to the head at a consistent and selected rate. Also the head may not immediately feed the ball [other heads may be feeding, or the user may have selected the feeding to be off] - this requires a place in the head where the next ball can be 'parked' before feeding - this is actually more complicated than it sounds as the head may be tilted, moving, or even be upside down!.

After a bit of thought and plenty of mucking about I came up with a relatively simple design that queued the ball into a place behind the wheels that fire it out, and park it. A servo arm then feeds it to the head when required - the movement of the arm pulls the next ball in from the feed of balls from the hopper and parks it. The only limiting factor here is the rate at which the servo arm can move - it is programmable, but it does have an upper speed limit - this limit will be the maximum feed rate the bot can do - not timed it, but it can feed a ball whilst the previous ball is still in flight, so I rekon its quick enough.

The next thing to do is to get round the odd ball being jammed and not being fed properly - sometimes when the head is at certain angles, the ball doesnt park correctly and the feed, the arm still tries to feed the next ball into the parking spot and the machine jams - I need to get the shape/size of the parking area just right so that the feeding is fluid whilst the head can be at any angle [nearly there!].

Another major problem, but hopefully not a show stopper!! - is getting the correct power fed to the servos and motors - At the moment there are several power feeds required -

xyz 3x servo head control -> 1 x 6v supply [currently a battery pack] or 4v

2x spinning wheels -> 1 x 12 v supply [mains via a transformer] 1 servo feeding/parking the balls 1 x 6v supply [battery pack] or 4v

1 servo stirring the balls in the hopper [may not be required, but is in the design] 1 x 6v feed. [or 4v]

I want the whole thing to work from a single plug socket plugged into the mains. To do this requires a multi-feed transformer - the nearest thing I can put my hands on is an old PC power supply - this has a mix of 3.5v, 5v and 12v switched supplies - This will be fine for the motors, but not for the servos as the power they require is very precise [give them more or less and they either dont work, or jitter really badly] - so any suggestions as to how I knock up a multi feed 12v, 6.5v and 4v supply would be greatly appreciated.

BS.

Just a quick update on the bot, still waiting for some materials [mainly some more powerful 12v motors]. However had a real flash of inspiration last night on one feature that would be very desireable. I'm going to integrate a wireless remote control unit [got one already] where I can click a button on it and it will immediately feed the ball I want.

This doesnt sound like much, however it will enable me to get a ball fed at the correct 'time' inline with my stroke. I wont have to tell the bot how fast to feed me because I'll get a ball when I think the timing is right - so how will this help? - well, when practicing my serving and third ball - I can serve, click the button on the remote [with my free hand] at the point I think the opponent would have hit the ball - the bot then feeds me a ball [spin and placement depending on what button on the remote I press and what shot is encoded to it] which I then play - this keeps the practice fluid and the timings feeling natural and not forced. Also when practicing rallying, I can time the button press to the point where the ball I've hit its the imaginary bat of the opponent  - so whilst playing the rally I can speed up, slow down and play the ball pretty much to the same timing as playing a real opponent.
I like this idea, it means I dont have to pick an arbitrary feed rate for the balls, they will come when I decide in real time, [although the usual automated rate would still be an option].
BS.

Ok, in the last blog, I identified the core systems that need building.

1. Ball container

2. Feeding

3. Aiming

4. Launcher

5. Ball Recycler

6. Software/Control

Here is a brief [somewhat brief ;0) ] description of each part....

1. Obvioulsly the easiest thing to design & build - just a small plastic bin with the botton cut out - at a later date i'll pop a ball sorting mechanism in there to stir the balls and stop any potential jams.

2. This one in tricky - I want the Robot to feed balls from two independent launcher heads - this basically means having a device that can selectively put a ball into either head, or just the same head, or in a sequence [say two to head #1, one to head #2]. A solution to this was to have a servo which literally 'pushes' balls to a left feeder or right feeder -which works under software control. The reason for two heads allows the system to supply balls from one head, whilst the other spins up to speed - keeping the flow of balls more or less continuous whilst the spin/speed/direction of the balls changes.

3.Aiming - This one was the main problems, it took the longest time to get to a solution that was both cheap and flexible enough to handle what I wanted i.e. Pan, Tilt, Roll, fine degree tuning, need to move at least 1-2 pounds in weight [this introduced another problem - as the wheels are spinning on the launcher they have a natural resistance to movement - this meant stronger [and more costly] servos - Currently there are three servos handling this [per head], each with 15kg stalling torque - so they are quite powerful.
They each move to quite a fine degree and vibration [another pain!] is kept to a minimum.

4. Launcher - this is the specific part of the system that spins the balls and throws them out to the player. I have originally designed a system that [believe it or not!] didn't use spinning wheels [as I believe ALL TT robots do at the moment] - However it became apparent that it was going to be costly and required more precise engineering, the plans are still viable and may be used at a later date however.
Currently I have two wheels that spin independently for each head - the relative speeds of each wheel controls the spin and speed on the ball - This can produce heavy spin, no spin, balls literally dropping out of the machine, to balls flying at, [last time i used it], 20+feet. The wheels are soft rubber [not sure if this is the best material!] - and are driven by 6v dc motors @ appx 15000 rpm - If I change anything in the near future I'll change the Dc motors for something with more torque and lower rpm's - the ideal solution would be to build in a clutch & gearing system, but, again, cost is an issue and I dont fancy designing a gearbox!. The circuit board that controls each spinning wheel uses PWM to set the speed - this is a system where the same power is always supplied to the wheel [to maintain torque], but the power is supplied in pulses - more more pulses, the faster the wheel spins.

5. Only part of the project i've not really worked on - I know what to do, but the robot can be up and running feeding balls from a large bin before this needs to be developed - so this will be done after the main project is finished.

6. This is basically a laptop and some software to control the systems. As a shiney new laptop would bust the budget - I'm using an old one I already have. I have some circuit boards to interface the laptop via a usb cable to the robot - This allows software on the laptop to individually control 10 servos, 4 dc motors and 10 led lights/5v switches. The cost of the boards were the single most expensive part of the project and came in at around £60.
The software i'm using is PYTHON - its fast, free, and can interface with the circuit boards with no problems - There is a 'speech' recognition module for Python that can 'listen' to vocal commands and control the robot. I'm currently writing the software [VERY time consuming] - and development will take a good while - although it is ticking along.

NOTE: At no point have i used a soldering iron, all of the circuit boards, materials, motors and servos are ALL off the shelf components :) In the next few blogs i'll go into more detail on each unit and will [hopefully] show some images/videos. I also want to mention that i'm open for suggestions, tips e.t.c. an any part of the system. So if you have any ideas, let me know.

BS.

Well, as mentioned, I've been working on this, on and off, for around a year or so now - The main problems with development have not been technical, but have been more to do with fitting time in to do it.

The main technical 'snags' have been more to do with picking the correct materials, engineering stuff [i've no electrical or material engineering background] and cost. Cost is a big issue! - Its forcing me to fabricate stuff myself instead of buying it - and mistakes are very costly as I have to purchase more materials and corerct my mistakes - an engineering background would have helped lots!!! [software engineering is basically programming and has nothing to do with Engineering] .

So what state is the project in? - Well I rekon I could build an amazing robot for around £2k :) - however only having a rapidly runnning out budget of £150ish i've had to make some large compromises. - despite that, the list below shows what I have actually done so far - My main focus was to take the independent parts of a robot, split them out and try to make a solution for each part - then [the most challenging bit!] - engineering them down to work with each other.

I came up with the following main parts of a TT robot.

#1 - Ball Storage - The main container for the balls.

#2 - The feeding mechanism - The device/system of getting the balls from the ball container to the launching device.

#3 - The Aiming mechanism - The device/system that pan/tilts the launching device in a way that allows it to lauch balls at certain parts of the table.

#4 - The launching device - A system/mechanism that propells balls with variable [or no] spins at different speeds and feed rates.

#5 - The ball recycling device - A system that collects returned balls and feeds them back to the ball storage device.

#6 - The 'Brain' - A system tha coordinates the appropriate devices to feed the required sequence of shots. This device also responds to vocal commands and reacts in the appropriate manner. 

I have built/prototyped mechanisms to do all of the above in one way or another - This proves, to me at least, that the project is viable and that a suitable robot can be built [budget notwithstanding!]! - In the next blog i'll detail what I have done in each case - i.e what my solution was!.

I just want to add that I'm building this for my own use, its not a commercial project and its just for fun.

BS.