ROV Driving, Localizing, and Data Collection

The 2012 Cal Poly ICEX Team

Dr. Clark visited us from Princeton over the weekend and we had an opportunity to hit the pool and play with the ROV! I have to admit, I was pretty surprised by how small it is; certainly beneficial for navigating the cisterns! We went through everything involved in preparing the robot, setting up a work site, and driving around underwater.

We had some difficulties connecting the sonar module, as the Serial - USB converter buffered data and disrupted the data stream. After a bit of fiddling with the sensor and connections, we found a laptop that still had a RS-232 port built in. This eliminated the buffer problem and was able to give us full sonar data.

Each member of the group took turns navigating the Poly Canyon swimming pool with our backs turned, using only the control box. This was a pretty good simulation of what it will be like to operate the 'bot in a cistern environment. After everyone was familiar with the basic controls, we threw some dive toys in the water and practiced retrieving our 'artifacts' with the robot gripper. It was quite a challenge!

Unfortunately, one of the drive motors had suffered water damage at some time in the past. After a few hours of running, the motor failed and bound up, cutting our session short. It took us sometime to trace the problem down to the motor, but when we finally pulled it out it was unmistakably caked with salt corrosion. We'll have to wait a little while to receive a replacement, but at least we all got a chance to see and drive the 'bot!

As part of our weekly lab meetings, so far we have heard from Billy and Jeff. Billy gave a presentation on the workings of FastSLAM and how the robot localizes itself while mapping out a cistern. In addition, Jeff gave a presentation on how this mapping data is processed to remove holes and create a traditional .m mesh file describing the geometry of the cistern. From here, I will take the mesh file and perform projective texturing onto the surfaces using the pictures gathered from the ROV. Now that we have an understanding of localization and mesh generation, Dr. Wood, Billy, Jeff, Tim, and I will be meeting sometime in the next few days to discuss the algorithms we intend to use to perform the texturing and bump mapping of the cistern.