Days 1 - 7
Days 8 - 16
Days 17 - 27
Days 28 - 34
This week I took 3 days off work to concentrate and getting the robot done for the competition in only 7 days, holy crap!! Robot construction is coming along well. I have all the sensors mounted except for the camera. The wiring is almost done, I have only the ultrasonic sensors left to compete. The wiring is a bit of a rats nest, but with the help of about 2 dozen tie straps I think I have it under control. I do still have to hack together an interface board for the ultrasonic sensors. basically just 6 voltage dividers.
The software is coming around nicely. All the low level drivers have been done and tested. The motor control subsystem is complete and tested. It should be noted that I modified the system design/block diagram. There is no longer a separate MSP430G2553 for the executive function. A single MSP430G2553 handles the compass, GPS, ultrasonics, and executive functionality. The new block diagram is shown below.
Note there are still three MSP430G2553's in the robot. The third MSP430 is used to control the pan of the camera, and interface with the host netbook running the computer vision system.
The computer vision system is shown below. It uses a netbook running WindowsXP and the Robot Vision Toolkit (RVT). I bounced around the idea of using a Rasberry Pi, BeagleBone, and even a Chumby as the CV host. The netbook was a last resort simply because I am running out of time.
The vision system simply outputs a 1 through 9 to indicate if it sees a cone, and in what region of the image it sees the cone. The MSP430 handles creating the RC pwm signal for the pan servo, panning the RC servos, and listening to the vision system over serial. When a cone is detected, the vision system sends a 1-9 to indicate in which region of the image the cone was seen. The MSP430 takes the region and the current pan servo angle and calculates a bearing. The bearing is sent to the Executive via a pwm signal.