Hello! This is Jamie. I’m a student at Olin College of Engineering, taking a class named “Iterate” where we research entrepreneurial opportunities in pursuit of our curiosity. I was wondering if you would be a good group to ask some questions about the robot localization app that I was trying to implement.
The basic form of the app is an independent localization app on a standard android device, which utilizes onboard sensors such as the IMU, GPS, and the monocular camera to provide some well-characterized information regarding the current position of the mobile platform.
In pursuit of clarifying some of the questions I had about the project, I drafted a series of questions that might help answer the assumptions I had about the value this project would provide.
In line with that, I’d really appreciate it if you could fill out the google form! If you don’t have time to take the survey, or would prefer to have a more in-depth / personal conversation, please contact me over email at [email protected], or any other means that feels most convenient for you.
Thank you!
Feel free to raise questions here which people might answer more openly (discussion).
In retrospect, the survey was probably not the best idea to inspire open discussion in the community. What do you think about the general idea?
The idea is fine, and has likely already been done and integrated into existing apps. For example, find an app (of which there are many( which allows you to get the dimensions of a room by positioning corners using the phone’s camera. The app makes use of the IMU and the camera, but not the GPS since it’s too imprecise. The issue is using the camera to “recognize” locations in the room, which is more SLAM or 3D mapping. GPS signals cannot really penetrate most buildings. An idea would be to somehow position a high resolution camera at the ground and check the positioning of pixels and patterns (like an optical mouse, but with many more pixels) to use to get displacement. You would then use a filter (Kalman for example) with the input from the IMU and camera to try to decently track position. You’d still likely get (positional) drift issues though.
