I am Apoorv Goyal.I am currently an undergrad majoring in a Electronics and Comms. Engineering.Because of a competition I came across flight controller Pixhawk and consecutively ArduPilot.The competetion is based on certain tasks that are to be undertaken autonomously without human interventioncomp link.
Major modules in our system are:
- Telemetry and Communication(Video Feed)
- Camera(Object detection)
- Companion Computer(Image Processing)
- Flight Controller and GCS
- Dropping Mechanism
- Autonomous maneuvering through obstacles(still and moving).
- Detecting objects on ground while in air.
- Detect person from air and drop a package.
- Telemetry logging
I would like to move on with the planning part so what i propose for GSoC is planning path through obstacles using RRT (Rapidly-exploring Random Tree) running on a companion computer .Also i would like to continue with using this module to plan paths through areas with Stay Out/No Fly zones considering them as obstacles and using this algo to determine path.
Some doubts that i had in mind were ,if it would be better to use the algorithm on a companion computer on the drone itself(continuing with mission even if the GCS link breaks) or using the algo to determining way points on GCS and send the calculated path over the flight controller(eliminating the need for an on-board companion computer).What I inferred was that it was highly dependent on nature of mission.But still some performance regarding clarifications would help.
If you have any suggestion,reach me through email@example.com. Thank for your time.