Overview

As we head into the Fall 2023 term, the Autonomy team will need to start integrating the computer vision modules needed for auto landing. Currently, we have our system get images from the video_input worker and detect landing pads in the image with the detect_target worker.

In the Fall term we will start to integrate more of the modules we worked on in the Summer for auto-landing. A diagram of the pipeline is shown below.

The video_input_worker and detect_target_worker have been flight tested and are confirmed to work.

The flight_input_worker and cluster_estimation_worker are currently being developed. PR’s have been made for both of these tasks and needs to be reviewed and fixed. Along with the code for these workers, the members have also written unit tests for their modules. These need to be merged into the codebase

Note: For the cluster_estimation_worker we may want to put a threshold on confidence for the detection so that we only take detections that are very likely to exist.

The data_merge_worker and geolocation_worker have been merged into the codebase and have corresponding unit tests. We will need to test these workers on the drone to confirm they work.

The controller_worker has been merged into the codebase and has corresponding unit tests. For the Fall term, we want to take the simplest path to achieve auto-landing. As a result, we will have to make a new controller_worker that is much simpler.

No work has been done on the command_output_worker . This will need to be a task that is done during the Fall.

Flight Test Plan