A computer vision dump-truck robot for IDDFA19 at Cornell Tech

Matt Freed and Bil Leon

Rough form:

https://www.youtube.com/watch?v=ux81tY8T8iU

https://www.youtube.com/watch?v=gVR7a9DyGOU

https://www.youtube.com/watch?v=SawVYpYz-DI

2 Motors

Caster Wheels

Battery

Camera

2x Arduino

Arduino Motor Shield

RPI

Servo

Mostly everything will be located under the hood. The camera will be located on the rod as seen in the renders

Shop bot is a workshop-helper. No more needing too put your tools in the right locations, simply put the tool into shop-bots basket and it will bring the tool to the right place!

Download the cad files that are located in this repository. These parts were custom designed.

These parts were designed to be 3d printed. 3D print the files and assemble them, referencing the render models for where each part goes. 3D printers are not exact when printing parts, and therefore the parts may need be sanded down to freely move depending on which 3D printer was used.

Once all are attached, begin mounting the elctronics. The servo belongs in the front, and should friction fit in place in its locking location. The motors also should friction fit into place on the rear of the robot.

Screw 2 caster wheels to the front of the underside of the robot on the platforms supplied.

Attach the arduino with the motorshield to the underside of the robot, located near center. Connect the motors to the motor shield.

Once the entire robot is assembled, place the second arduino and the Pi into the electronics bay (the fin-like box).

Connect the servo wires to the arduino located inside the electronics bay.

Connect the two arduinos now to the Pi's usb ports, along with the camera usb.

Flash the servo arduino with the servo code and the motor arduino with the motor controller code.

The servo code is a simple "sweep" code adjustment, where the servo will move to 1 position, then return to its original position.

The motor controller code utilizes a library which allows the motors to run both forwards and backwards. This is necessary for moving backwards and turning.

Flash the Pi with the supplied code for computer vision, this will tell the pi what is inside the dump bed, along with what arduino to send information to. The included TensorFlow Lite model includes detection for hotdogs (and aggressive detection of humans).

Power the robot with a USB-C Power Delivary battery and place a hot dog inside the bed! If done correctly, the robot should see the hot dog, beep loudly, and know that it must go straight, then to the right, then dump the hotdog, and then return back to you!

Congrats you are now a robot wizard, harry!