from: 10Jun98 at 9:00am-11:00am
with: Dale, Samer, Jack, David, Alex and Martin Stonebridge
Meet Martin Stonebridge
Discuss camera enclosures
Discuss end cap extension
Discuss mounting brackets
Status from everyone
Review our submission to the OCEANS98 conference (incomplete)
Vaganay, J., Jouvencel, B., "Motion Estimation for ROV Stabilization with a Light-stripe Sensor," Proceedings of OCEANS96, Fort Lauderdale, Florida, USA, vol. 12., pgs. 1088-1092.
Review proposed software architecture
Jack has been working on designs for the camera enclosures. He implemented image copy code for the FeatureTracker module. Dale spoke with Jochen about Kalman Filter C++ code. It looks like a poor implementation so we will likely write our own for the StateEstimator module. He helped Chris make a veriboard to fit inside the motor and packed up the KVH to return to Wollongong. Chris continues to write base classes for use on various modules. He has prototyped an encoder to fit inside the motor with sufficient capability to accomondate shaft motion. David coded some parts of the SensorSampler module.
The computing design has been reviewed by several external experts. The strongest recommendation was to use something other that LM629-based boards for motor control with the suggestion being a single MC68332 board. This looks like a good option. However, since the MC68332 is a Motorola board the option of using a PowerPC chip as the main processor instead of Intel Pentium is being considered. The main issues are availability in either CompactPCI or passive-backplane PCI and cost of the necessary VxWorks kernel.
We had very fruitful discussions with Martin Stonebridge about making parts. The camera enclosures are quite mature and will soon be ready for fabrication. Design of a new end-cap needs to be detailed but seems quite possible. Martin also had a number of suggestions for fitting bits inside the motor enclosures and indicated fabrication of heat sinks would be possible.
We briefly reviewed the paper and found some useful insights into Kalman filtering inertial sensors. We also talked for a few moments about the software architecture. This cursory introduction will warrant further discussion as software implementation continues.