The following entry is adapted from an email from Dominic Jonak, Senior Research Programmer on the DepthX project.
We deployed the robot down the center of Verde and returned to double check the DVL [Doppler Velocity Log]. This procedure worked well. Unfortunately we continued to have trouble with the SLAM [Simultaneous Localization and Mapping] system.
As a result, much of the day was spent with the robot operating autonomously [without human intervention]. Dominic Jonak created a randomized exploration strategy that was intended to make the robot “bounce off of walls.” Unfortunately, the robot did not “bounce off of the walls.” Instead the robot spent most of it’s time in the center of the cenote due to sonar noise that made the robot think that the walls were a lot nearer than they actually were.
Even with these problems, the randomized procedures resulted in good coverage of the cenote walls that enabled us to produce some nice maps.
A map of Verde