This week was spent finalizing designs for our system. Some parts have been acquired including aluminum tubing, corner brackets, wheels, and a lazy susan bearing assembly. The gripper had been prototyped in Lego and is currently awaiting a 3D printed chassis. Preliminary computer vision has been tested and can identify blocks from a distance somewhat reliably given ideal lighting.
The computer vision prototype was a first attempt at locating which blocks exist and which blocks are missing. The basic image flow involves blurring to reduce image noise, edge detection and finding contours of block shapes. Finally dots are rendered at the center of each contour.
This prototype works fairly well as seen in the video, but could use some improvement to fix errors around wood grain and shadow.
We will be using a square extruded aluminum tubing for all but the vertical axis movement. The tubing was purchased from Burcham's and has a consistent width of 16mm on the outside and about 13mm on the inside. We will largely use 3D printed mounts for our frame to maintain a clean and stylish look.
For our corner mounting, we wanted a flush mount. Each bracket inserts into the aluminum tubing and has holes intended for optional extra support. The brackets were 3D printed in order to get this exact shape. We are very happy with how they turned out and how seamless our corners look.
We have purchased the bearing assembly for our rotating platform to which we will mount our gripper and poker assemblies. We will also have a motor for rotating the platform, controllers for the motors and servos, a camera for precision computer vision, and possible beam break sensors for accurately detecting missing blocks.
The current prototype is made from Legos, a 3D printed base, and laser cut gears. The gripper began as a Lego prototype with a simple parallel gripping mechanism. In the second iteration, it was recommended that a differential be added in between the driving gears to assist in fluid grasping and force control. With this addition, it became apparent that the Lego mechanism was incredibly useful, and we wanted to keep it in the final design instead of purchasing more complicated and expensive parts. However, most of the Lego prototype was purely for stability, and even then did not provide as much support as we needed. Thus in the next iterations, we opted to replace the stabilizing Legos with a 3D printed base, and keep the Lego grasping mechanism and the differential for simplicity. Three versions of the gripper base have been printed as of week 4, each one an improvement on the last.
We also demonstrated servo motion using an Arduino UNO. The servo has 3 states which correlate to the open, partially closed, and closed states of the gripper. The gear on the servo fits with the Lego gears on the gripper prototype.