Robocademy Field Trials in Scotland
Training Session in Loch Earn
The Eurathlon 2015 Challenge is already casting its shadows and our Robocademy ESRs are constantly working and training in their teams in preparation of this event. The unofficial Robocademy Team “Nessie” (Heriot-Watt, DFKI, Atlas, Seebyte) performed some field trials between the 18th and 20th of August at the Loch Earn Lake in Scotland. ESR4 Mariela de Lucas provided us the following report and pictures of this Event.
Our goals for the day were:
1) To test the comms between Nessie and EMILY in order to assess the operability in an untethered mission
2) Run the re-installed cameras on Nessie to gather image aquisition of the pipe and valve dummies for the purpose of training the object detection algorithms.
We arrived to the site at around 11 am and started by setting up our equipment. The day was sunny and this allowed us to work comfortably in the dock and grounds. After lunch, we deployed the valve and pipe dummies in the loch and were able to obtain a large amount of images. We did encounter an issue with the usage of the cameras and found out we cannot use them all simultaneously due to a limitation of the USB camera driver. We have found a workaround and are planning permanent fixes.
Unfortunately, we were unable to complete the test of the comms due to some issues with the WiFi router and we will attempt to test this tomorrow.
In recap we can call the events of the day a success. Weather forecast for tomorrow does not look very promising but we will nevertheless do our best to get the camera calibration done before the rain comes.
Our goals for the day were:
1) Test the comms between Nessie and EMILY (body requests, mission status and command)
2) Test path planning executions
3) Camera calibration
4) Resume the acquisition of images of OPIs (numbered buoys)
We were able to work in dry conditions all morning during which we simultaneously tested the comms on EMILY and gathered images from the numbered buoy (which we missed yesterday) and additional footage of the valve and pipe. EMILY kept loosing wifi connection and we had to resort to fixing the modem to the dock. We tested the modem stand-alone functionalities and were able to do some tuning on the controller for EMILY separately. We also recorded some nav information to examine the precision of the IMU sensor.
We ended this part of testing just before lunch and found out that Nessie's second computer was not booting. It was because of this that we could not perform the calibration of the cameras as these are connected to this computer.
Fortunately Nessie's first computer was enough for testing the path execution. After lunch, rain caught us mid-experiment in the boat but were able to finish testing the lawnmower and OPI inspection patterns.
All that is left to test tomorrow are the autonomous features of a state machine that we've built specifically for these trials. This will allow us to test the changes of the behaviour in the current environment conditions.
Our goals of the day were:
1) Test the autonomous features of a state machine
2) Resume tests on EMILY by sending waypoints and observing the controller behaviour
Subsequently, tests included only EMILY and continued until around 16:00. The modem capabilities were now tested in open water where the connectivity seemed to improve.
The upcoming weeks will require a great deal of software integration with the state machine. In addition, we will focus our efforts into finding the problem with the second computer on Nessie and finding a permanent solution in time for shipping the vehicles to Italy.