Future Project Ideas
Here we will go into more depth on what each of these projects would require:
Migrating Swarm Client PID's to Firmware
Focus Area
Firmware Programming and Communication Protocols
- This project requires heavy modification of the Firmware and Communication systems inside the Crazyflie.
Tasks
Firmware
- (HARD) Modifying Commander Packet Structure to allow position data to be sent rather than pitch, roll, yaw, and thrust setpoints.
- Modifying Firmware to pass the new position data into the correct place for use by the Firmware PID controllers.
Client
- Modify Swarm Client packet construction to match new commander packet protocol in Firmware
- (EASY) Remove Controllers from callbacks and just send location packets to Crazyflies while still maintaining other functionality.
Benefits
- Significantly reduces Client-side computation and allows Crazyflie swarm to share the load.
- Potential extension: Completely independent Crazyflie flight (i.e. No Client needed at all)
Model Development/Verification and Tuning X, Y, and Z Location PID's
Focus Area
Control Systems and Model Development/Verification
- This project will require developing a model of the Crazyflie system and using the Model to calculate PID constants.
Tasks
- (HARD) Identify model Parameters for the Crazyflie
- Construct model in Simulink and Verify simulation compared to test data
- (EASY) Derive PID constants from verified model, test flight performance
Benefits
- VERIFIED MODEL!! We can do many things once we have a verified model of the system.
- Flight performance will be greatly improved allowing more complex applications to be developed.
- Potential extension: Path prediction algorithms and optimal pathing between points with obstacles
Adding Parameter Support to Swarm Client
Focus Area
Communications Protocols and Networking
- This project will require Client-side software development to create helper functions for easy modification of parameters on the Crazyflie Firmware
Tasks
- Creating packet framework in the Client to replicate the parameters packet protocol
- Programming helper functions to make it simple to modify parameters
- Ex. Write Parameter Value:
setParam(string "param_group.param_name", int set_value); - Ex. Read Parameter Value:
readParam(string "param_group.param_name");
- Ex. Write Parameter Value:
Benefits
- Won't have to re-flash Firmware every time we want to test a parameter change (Improves Firmware testing efficiency)
- Less Firmware modification if we can just write params during Client init.
- Potential Extension: Tuning Firmware PID constants
Threading Each USB Radio to speed up Computation
Focus Area
Computer Architecture and Parallelization
- This project will require/develop a good understanding of parallel processing and computer architecture.
Tasks
- Apply threading to each USB Radio individually
- A single Radio will be assigned at most 3 Crazyflies, and each thread will handle the corresponding Callbacks for those Crazyflies (which includes control calculations, Radio communications, and logging for each Crazyflie)
- Troubleshoot any issues that come up with non-sequential program execution (shouldn't have problems)
Benefits
- Greatly reduces time required to complete a loop cycle (relative to number of Crazyflies in the Swarm)
- Improves scalability of the Swarm since we don't need to worry about missing the Camera System new data timing deadline as the Swarm size increases
- Potential Extension: Increasing Swarm size to Camera System capacity limit.
Developing New Controllers for the Crazyflie (pending Model Development)
Focus Area
Control System Development and Implementation
- This project will require a background in theoretical controls modeling and systems design.
Tasks
- Research Control Designs and decide which controller you want to use
- Examples: State Feedback Control (LQR, LQI, Pole Placement), State Space Controller (Lyapunov Estimator, H2), etc..
- Develop a Simulink model for the controller and implement design in C++.
- Verify model by comparing to real-world tests on the Crazyflie Swarm!
Benefits
- Improved flight control leading to more complex applications
- Maintain control without all the information (State Estimation)
- Potential Extension: Stable control with Stochastic lossy packet transmission