25th Mar 2017
Starting to feel the time pressure a little now…
TITO 2 has progressed from early prototype into a the early stages of final bot. This means that the laser cut chassis has been altered to include all known mounting points for the electronics as well as shaped for the b-e-a-utiful 3D printed clam shell bodywork.
What this has meant is that the motors, battery, wiring harness, pi, connected sensors and all other items have been meticulously and carefully transferred from the prototype to the release candidate chassis. The result of that is the motors are now obviously backwards, the left distance sensor is now the front…well, you can see where I’m going. The motors being wired backwards was actually a blessing in disguise as it wasn’t noticed till then that RC mode was driving backwards (sad face).
So, RC mode now correctly drives forwards, but our autonomous modes all flip out. It should be a simple matter of code changes, but it’s all time that we know we have little of.
The team are working tirelessly to get the robot ‘fighting fit’, so Paul has taken the I2C circuit design and strip board away to solder up a proper (non breadboard) I2C connector/extender for all the I2C sensors and screens that all need to connect to the single set of 4 pins on the Pi. Martin, Rob and I have gone to sit in a dark room and think about what needs to be done on the code base to progress the project ready for yet more driving tests and tweaks on Monday. Paul also knocked up a laser cut rig for the golf challenge, so that will need to be assembled and ‘electrified’ ready for testing on Monday as well.
Mark has been busy in the evenings designing 3D objects to be printed that will safely hold our sensors so that the electronics don’t fall off or get damaged by inevitable contact with the walls.
Brian has volunteered to take charge of the line following sensor. This is an area that has hit problems. We noticed that the line detection was sketchy at best when using the line sample kindly provided by Piwars Mike and Tim. We think the sample has been laminated with some sort of IR reflective coating (probably to stop the course/banner from fading in sunlight). This means that our sensors (and apparently other groups sensors as well) have struggled to see the change in colour as they only see a reflection. The effect is similar to the 2 way mirrors seen in films where several ‘usual suspects’ line up in a well lit room. They cannot see the people judging them in the dark room the other side of a pane of glass, mainly due to their rooms reflections overpowering the small amount of light coming from the dark room. To try and overcome this issue we have been playing with all kinds of techniques like blocking out all external light sources, changing sensor distance from surface, angle of sensor to surface, disabling onboard IR LEDs and using an external IR source at a significant angle difference. None of these bore much fruit and neither did the following test of seeing if out IR sensors were sensitive to visible light. Good news, they are…bad news, not enough (again, sad face). Our final tests was to use black electrical tape over the top of the sample line. This instantly was a solid detection using the standard sensors in the standard manor. Our options are rapidly disappearing. Two remain, use a different senor to detect the line (probably the pi camera currently being tested by Brian), or kindly request Piwars overlay black electrical tape onto the course.
Anyway, to quote a good film ‘Keep moving forward’ (without googling, you have 3 guesses which film).