Today we assembled the "kitbot." Evan worked on CAD, mostly on the drivetrain, but it is very far from being finished. We need to decide on a manipulator. Hopefully tomorrow we will be able to drive the robot around with arrival of some missing components.
Andrei: playing with the tablet. Problems with Windows installing Maple drivers, decided to go with Ubuntu.
Andrei: Installed Ubuntu, made the kitbot work. Started figuring out OpenCV basics.
Evan: Started designing.
Andrei: Problems with OpenCV on the tablet. "make" freezes the tablet. Had to reinstall Ubuntu. Seems to work now. Vision got seems to be in a good shape
Evan: Had major discussion on strategy and plans. Continued designing.
Andrei: Tablet works with OpenCV. Made the kitbot follow the wall.
Evan: Modified our mechanical strategy significantly to better match our game plan. Lots to redesign..
Andrei: Decided to go hardcore. Made a simulator of a real world. Now don't really need a kitbot for a lot of stuff.
Evan: Started designing ball collection implementation - sweeper type collector.
Andrei: Started implementing the state machine. Takes a long time...
Evan: Continued designing ball collector.
Andrei: Not too much. Some Wall Following stuff.
Evan: Made prototype Archimedes screw.
Andrei: Coded path planning, reading stuff about localization
Evan: Tried to sort out the rest of the general geometry of the robot. ~85% designed
Andrei: Working on localization. Much problem, such pain.
Evan: Added on board ball manipulation hardware for after the balls are collected.
Andrei: Localization kinda finished. Painful process of putting everything together, but seems to be working more or less.
Evan: Trying to sort out mating details, etc.
Andrei: Getting some rest. Evan: Really trying to finish designing. Pretty tired though, so it will wait until tomorrow.
Andrei: Trying to get response from sensors. Many deaths. Sad. Evan: Machined a couple interface parts for the Archimedes screw. Finalized first level design. Ready to start laser cutting tomorrow.
Andrei: Spent the whole day figuring out what to do with range sensors. IRs definetely suck in terms of range and accuracy. Sonars suck for their death count. We are thinking of buying more expensive sonars.. we'll see. Evan: Took a trip to home depot for some PVC pipe and round stock. Finally started building. Found out that tolerancing slots is pretty unnecessary; should have found that out earlier.
Andrei: Waiting for the real robot. Realized that the localization is not realiable at all, found many bugs.. Still sorting stuff out.