[Dprglist] RoboColumbus waypoints

[David Anderson]

Howdy DPRG,

Interesting discussion at Saturday's DPRG (virtual) monthly meeting 
about the upcoming RoboColumbus and some of the challenges involved in 
doing it virtually/remotely.  Doug did an overview of the contest and 
current rules and provided some possible modifications for the online 
event scheduled to happen next August.  I'd like to offer a more 
generalized suggestion.

As Doug described, the contest is designed to get progressively more 
difficult.  For example, the first leg, from the origin to the first 
cone, is a line-of-sight straight line with no obstacles.   The second 
leg, from the first cone to the second, includes some undefined group(s) 
of intervening obstacles: trees, picnic tables, and what have you.  It 
is intended to demonstrate if the robot is capable of doing obstacle 
detection and avoidance while navigating to the next cone.

With the current rules, the roboticist is allowed to walk the contest 
course with his own GPS and record as many data points as needed to 
navigate around the obstacles, and then download those to the robot 
before its run.    You must admit that in that case, it is the human 
doing the obstacle detection and avoidance and path planning.  Not the 
robot.

Nothing wrong with that.  And it has it's own challenges.  But it 
doesn't seem like it is the same contest.

Perhaps we need a separate category for human-planned vs robot-planned 
navigation?   In robot-planned, you only get to use the waypoints handed 
out by the contest committee.  No extra obstacle-avoiding waypoints 
allowed.  In human-planned you can use as many as you want.


just some thoughts.

cheers!
dpa