Tag Archives: Autonomous

Folkrace 187

Posted on by


In Swedish Robot Championship there’s a tournament that is called Folkrace where autonomous cars is racing around a track. A sub class of this is a miniature race in scale 1:87.


So I started out on a Fuller Car system truck in scale 1:87. It have some really nice suspension and drivetrain but is made for following a iron track.


A NodeMCU development kit (ESP8266) almost fitted the truck bed, if i filed of 0.5mm of it i fitted perfectly.


I solved the steering by cutting a hole, big enough for a small servo, in the trucks chassis.
Then i glued on a small magnet that aligned with the already existing magnet in the steering system.


I then mounted 3 VL53L0X ToF (Time of FLight) distance sensors inside the cab of the truck. The sensors is pointing out from the same holes the windows are located.
The sensors will be used to locate walls and other trucks, it’s gives a pretty low resolution, but should be enough.


The space inside the truck bed is cramped, and I mean it. After fitting a small H-bridge, a couple of batteries and some cables together with a start module nothing more can fit.

As if the cramped space inside the truck bed was not a challenge big enough I choose to run Micro Python on the MCU.
It’s pretty nice and all, I could remotly upload new Python code to the flash and run my test scripts inside a Python interactive terminal.
You can find the code here: https://github.com/TimGremalm/Folkrace187

I found the challenges of the small form factor thrilling, it was a really fun adventure puzzling everything together and make it look pretty stock.
But there is room for many improvements! The Fuller Car system have a very nice steering system, and it’s very useful for a future design.
But Fullers drive train is a worm gear, it makes it strong but gives the drive train some momentum that makes the car slow in response when braking och switching between going forward and backwards. For future builds I would have to build my own drive train.

The cramped space inside the truck bed is due to a lot of premade modules and a lot of cables. A more effective way of doing it would be to make a PCB with a ESP8266, H-bridge and sensor bus built in.

Also I think I would abandon Micro Python for C and Free RTOS. The VL53L0X driver is very slow in Micro Python and it takes too long to read 3 sensors. The whole driver thing is pretty hard to fault find and gives great me a great hazzle.

Shooting aluminium with a large capacitor-bank

Posted on by

Påsse at CRF buildt a big capacitor-bank that could be charged with very much energy.
The capacitors are connected in series to a coil of very wide cable. When the energy is released to the coil it generates a gigantic magnetic field in a few milliseconds.


The  aluminum-disc are shot in the air. We calculated it flow about 20 meters straight up in the air.


The capacitor-bank is fired by Exploded.

Explodet – Automated fireworks-launcher

Posted on by

Background

Two days after christmas me and my friend, Andreas Svensson, were going to watch Avatar 3D, the tickets was sadly canceled due to SF the bitch. We then started to think of something else to do, Andreas: “..how about that fireworks-launcher we spoke of a few mounths back?..” And so it was decided.
We had a short deadline, roughly about 5 days.

Finished result

BTW. Explodet is short for Explosive Detonator.

Launch-day!

20:00 – At new years’s eve, five days of constant work, we finally made it. It was ready for testing.
21:59 – We arrived at the party, I had not slept for 36 hours
22:06 – I started programming at the party, Andreas went to pick up some equipment
22:25 – Party-cat Sigge was showing interest in Explodet, discrete sniffing and foxy glimts from the eyes was seen.
22:53 – Unfortunately the software wasn’t ready, still had hope I would program it outside
22:55 – We went out to setup all the gear
23:10 – The fireworks launching-ramp was rigged
23:21 – The 230V-generator and 2×500W halogen lights was rigged.
23:40 – Explodet is up and running
23:41 – I found out that it was to cold to program, -15 degrees C. We decided to launch the fireworks remotely (by pushing down spacebar) was good enough.
23:55 – Cable setup
24:00 – Still cable setup
00:01:01 – Success! First firework fuse is sparkling.
00:01:05 – Disaster! The firework won’t lift off, 20 peoples is throwing them selves down the hill.
00:01:30 – It all went well! 🙂
01:15 – The final fireworks went off in a beautiful sparkling rain.
01:16 – Starting to bring the rig down
04:00 – Finally home!

Specs. and parts

Research at Chalmers Pyrotekniska Förening

Before launch-day I went to visit Chalmers Pyrotekniska Förening. They showed me their fireworks-launching-case, it had 50 manually trigged outputs. They also had a sequencer with 25 automatically trigged outputs.

Building process

It was a perfect start of the project: when we first arrived a big christmas-tree made out of coke-cans met our eyes.

Just before the project had started Andreas said: “I’ve 20″ wide screen to no use at home, wouldn’t it be nice to mount it inside?”
I asked him how wide it was. “Roughly 245mm” I went to measure the case “It would fit, the case is 248mm!”
We started out by placing the 20″ screen inside the case. It was a perfect fit!

We marked 4 screws in the screen with paint and then placed the screen inside the case. We then drilled holes were the painted marks.

The next step was to mount something to hold the case-lid up.

We then tested if the screen was working properly, it did. Also notice the mini-computer placed inside the case.

Andreas i starting on the PVC-pipes to the launching-ramp.

I started to build the relay-card. First I mounted a 12V 1A voltage-regulator UA7815.

Flying coke-can in a stream of air.

First part-layout-testing. Battery, computer, interfaceboard, relayboard.

The inner case bottom plate is made out of a old computor-chassi-side. I measured it up and started cutting, then i had to bend the plate in order to fit all the switches.

Before painting, I polished the plate down to the steel.

I painted it firey orage!

A hole for the air-flow to the computer felt necessary.

To make the construktion sturdy, I manufactured some side-pieces.

In the middle of te 3:rd building-day Peter came around to fix his Guitarhero-controller. We then watched Monsters VS. Aliens an when Peter inserted the audio-cable into his computer he destroyed the christmas-coke-tree. We rebuild it again.

The key-switches mounted.

Second part-layout-testing.

230V enclosure

Case-bottom-plate holes for switches, air-filter, mounting of banana-plugs

I had to bypass the ordinary computer-power-switch and mount it on the case-bottom-plate.

Mounting of switches

Mounting signal-cables and attaching the relay-card

Will the plate fit? Yes indeed! We can even close the case without the switches touching the screen. Thighty!

Andreas is making Nikrothal-coils that will ignite the fuses.

I making a pair of new USB-connectors, the old one burned up! 🙂

The fireworks the bought from Chalmers Pyrotekniska Förening!

Installing Visual Studio 6.0

Close-up on the switches!

In the making of the launching-ramp.

Squeezing the cables nice together.

Erik and Erik came around early at the evening of new years eve to check on the progress.

The final touch on Explodet was the letters and the stickers. Behold!

Piratbilen V2

Posted on by

Background

The autonomous pirate-car was a huge success at Dreamhack W09, so Benjamin has built a new one. The Pirate-Car V2.

 

Parts

The chassi is from a RC-car, the stearing system has been modified with a servo and it has a better steering-radius. The motor and gearbox is original but it has a reflex-sensor built in the gearbox, the reflex-sensor is aimed to a gear that is painted half black. The sensor can pick up 3 peaks per revolution, these seams to be enough to make the PID-regulator stable.

It has five Sharp-sensors 3 in the front and 2 in the back.

The autonomous part

The Pirate-Car navigates by it self, it goes around tables, chairs, walls with no problems.
The two front sensors ,left and right, separately calculates how the car should turn. The two turn-values are then subtracted and makes a delta-value that is put out to the steering-servo. This works pretty well, in a corridor the car senses both walls and navigates right between the walls.

The Pirate-Car could sense if it’s stuck in several ways

  • Front-center-sensor senses a object close enough
  • Front-left and front-right sensors senses a object close enough at the same time
  • PID-regulator is delivering a high output that indicates a spinning wheel

 

If it’s stuck it starts the reversing-procedure, it reverse until the front-sensors feel it’s free or the back-sensors indicates a object.

Remote-possibility

The main-PCB has a IR-eye hooked up, this eye could pick up commands from IR-remote-controls. Benjamin has written a nice protocol for picking up commands.
Currently two commands could be sent:

  • Shut the H-bridge on/off
  • Increase/decrease the PID-regulators goal-value

With a Introkit09 we built a remote-control, we used the existing IR-diode to send out IR-commands. But the signal was weak, we then tied to switch to another type (TSAL6400) which were much brighter.

Future plans

  • Make a chaseing-mode, if someone/something tries to catch up the Pirate-Car it goes faster
  • When it slows down, brake-lights will glow
  • When someone tries to fuck with it, a loud annoying sound and flashing lights will go off
  • Following-mode, it tries to follow a person around

The autonomous part