375 Points | FTC Ultimate Goal | The Clueless

[@EliteCubingAlliance]

Hey, I saw your comment on a science fair video and was interested in what you had to offer. Your projects are truly amazing and you are highly underrated. I have a group working on a robot that is using machine learning to do object detection to find trash in a given area, kind of like the robot wall-e. We are thinking of taking a raspberry pi and a camera and using tensor flow to do object detection. We would give it an area, such as a parking lot, to look for things such as bags, boxes and bottles, then it would drive to those things, pick them up, locate the nearest trash bin, and throw it away. Do you have any suggestions for something like this?

[@sj2penrose716]

Do you guys use vision for auto aim?

[@theclueless11212]

no, we just calculate turret position based on the robot's position. We use the T265 for localization so we have very little and non-accumulating positional error.

[@siddharthrao6147]

@@theclueless11212 would I by any chance be able to take a look at your T265 Localizer class?

[@theclueless11212]

​@@siddharthrao6147 I'm not our expert on t265, but we developed all the JNI ourselves and I can tell you it took hundreds of hours and messy code, and the t265 only works under specific conditions. It wouldn't be practical to use in an in-person game. I would share our code but someone made our repo private so I can't.