Raspberry Pi is a great platform to be used in a robot project, but it has a severe limitation: by default it does not have a built-in power button. As a Linux-based system Raspberry Pi requires to be shut down gracefully before switchin off power. Simply plugging the power cord off may corrupt the file system on the SD card, which I have experienced too many times. To avoid corrupting the file system we constructed a simple circuit for shutting down power gracefully. The circuit also allows us to put our robot into sleep mode, automatically waking up the robot after specified time interval. Just out of curiosity we added possibility for measuring current consumption. Some commercial solutions such as Sleepy Pi with similar functionality do exist, but this kind of system is quite easy to construct from separate components.

Read more: When Everything Fails

Making a Raspberry Pi robot is surprisingly challenging. Sometimes the project advances in great leaps, then there are these days when virtually everything fails. Our navigation software is now mostly ready, so our project is just about to reach the primary goal we have set for this summer: to navigate autonomously on the local lake following pre-defined routes. We had planned a major test session on which we would test automatic navigation and celebrate the test result - which would obviously be a huge success. Because why not? Unfortunately this time all odds were against us.

Read more: Improving motor controller cooling
2 Microbot motor controllers with heat sink

During previous test drive the poor performance of propulsion system was identified to be caused by overheating of motor controllers. We did some experiments to measure how much current the motors actually draw when submerged in water. Cooling was improved by installing an aluminium heat sink. Motor controller performance was greatly improved, but a fan needs still to be installed to enable full power output.

On sunday 24th of August we went to local harbor for testing the Kort nozzles. The nozzles turned out to work as intended, but the overall performance of the propulsion was still surprisingly bad. Fortunately we managed to find out why.

Maneuvering the robot manually is now extremely easy due to a simple improvement made to the compass. During the test the GPS receiver failed to work properly, so we missed a good opportunity for testing autonomous navigation.