We are running Tensorflow in the FTC Blockly code based on the “example” in the robot controller app. The code works fine when it sees one orange ring for zone “B” or the stack of rings for zone “C” but when the Tensorflow routine returns a “none” (no rings = zone A) the code crashes and the driver station says there is an error at the call Telemetry.Update
We have an if/else/else that looks at the “recognitions” from Tensorflow. “Quad” comes back fine and the robot rolls to “C.” “Single” sends us to “B.” “None” crashes the whole code. When we bypass the Tensorflow and just use a controller to push buttons A, B or X ( C ), the robot runs fine.
We just used the Tensorflow Webcam Sample code in the app and then added our own code to create our autonomous.