The project team began their work on Robbie by writing out a list of goals for the final project.  This initial list included things like: obstacle avoidance, pink weed detection, blue weed detection, A-star path optimization, and other very base line tasks that would be necessary for the robot to function successfully.  Having identified all of their goals, the first section of code written focused on path following and obstacle avoidance without the help of A-star. This forced the project team to gain a better understanding of the LADAR angles: which values pointed which direction and how close of a threshold was too close before the robot would hit a wall.  The project team quickly learned that while this method was useful for future integration with A-star, Robbie would be much more successful navigating the given course with the help of A-star. Thus the next step was to implement A-star so that Robbie could be constantly updating where obstacles were in his line of vision using different LADAR readings and then add them to a map to help him optimally plan his path to the five different way points.

 

Once A-star was finally up and running the project team decided to tackle weed detection. This was done using a combination of a MATLAB function and Robbie’s camera.  The process began by determining the proper HSV values for both the pink and blue weeds, and then coding Robbie to tell him when 35 or more pixels of that color were visible he should turn and face them, and then move towards them until he is directly over them.  He then stops, “exterminates” the weed for one second, and finally continues on his path. The project team also included obstacle detection code from the work before A-star was added in to help with obstacle avoidance in weed detection. This way Robbie is almost guaranteed not to hit walls no matter what state he is in.

 

Satisfied with the way all of this was working, the project team began to look at how they could shorten time on their course run.  One feature that was added was when Robbie was outside of the course, moving to external waypoints, A-star was halted in favor of simple movement.  This allows Robbie to drive at an increased speed in a straight line to the desired waypoint, and then call A-star again when he re-enters the course. The project team also aimed to reduce time by only going to exterminate a weed if it had not already been visited. This was done by using distance calculations to determine the position of the weed when the robot turned to face it.  If Robbie calculated that he’d already been to that weed position, he skipped it and continues on his way.

 

Having finished all of this the project team turned to bells and whistles: what they could do to make their robot a little flashier.  The goal here was to include sounds, visual, etc. that would take extra time to code but make the robot that much more distinguishable from the rest.  This was all only attempted once the basics had been completed.