ap
Content
Development of an Autopilot for Small UAVs
Chinmay Patel, Aircraft Aerodynamics and Design Group (ADG) Stanford University
Abstract
This poster provides an overview of the development of a low cost autopilot for use in research and coursework related to small autonomous Unmanned Aerial Vehicles (UAVs). Important specifications and software modules are discussed along with a control law development framework. Flight test results are shown. The capabilities of the autopilot make it an ideal choice for small UAVs and related projects.
April 24, 2007
Specifications
The specifications for the 2nd generation autopilot are shown below:
• 29.5 MHz processor • High level C-like programming language • GPS unit with 4 Hz update rate • Two-way wireless link with ~1 mile range • Inbuilt manual override capability • Inbuilt servo PWM generator • Small size: 2.5 x 2.75 x 1.0 inch • Weight: 2.3 oz including sensors, battery pack, and wiring • Sensors tailored for small UAVs • Airspeed, barometric altitude, 6-axis IMU
Control Law Design
Control laws are designed using a non-linear dynamics model in Simulink®. A PID control example is shown here. Sensor and actuator characteristics are also modeled where necessary.
Capabilities and Future Work
Autonomous pursuit flight: (Video) One airplane, always under manual control, broadcasts its GPS location every second. A second airplane tries to autonomously follow the first airplane. Results with a very simple position feedback based control strategy are encouraging. Some of the capabilities of the autopilot are listed below: • A wide range of autonomous UAV missions, e.g., Autonomous thermal and gust soaring • Data logging, system ID and control law validation for unconventional aircraft • Multiple-UAV missions: • Formation flight • Collaborative search • Cooperative soaring • Convoy following and target tracking • Collision avoidance
Introduction
A low cost and lightweight autopilot is needed for a graduate course, AA241x, offered by Prof. Ilan Kroo and Prof. Juan Alonso. In only one quarter, student teams design, build, and fly UAV to fulfill a prescribed mission, using a GPS module as the only sensor. The ability to fabricate and modify the autopilot in-house is also needed.
Software Modules: Attitude Estimation
Body axes accelerations and angular rates measured by an IMU are used to estimate the attitude of the aircraft. An Extended Kalman Filter (EKF) based approach, shown below, is used. The rates are first integrated in time to get a predicted attitude. Accelerometer readings are used as measurements to correct the predicted value.
Results
Trajectory from a GPS waypoint navigation flight is shown below. This trajectory can be plotted in near real-time and also be integrated with terrain mapping software.
Conclusions
• The autopilot provides a lightweight and low cost subsystem for UAV related research at ADG and serves as a tool for coursework. • Students get valuable hands-on experience with in-field testing of UAVs, validation of control laws, and familiarize themselves with embedded electronics. • Since the autopilot is designed and assembled in the laboratory, it can be easily modified for project-specific needs. • Wireless communication links open up the possibility for research in multi-UAV systems.
For research projects, a lightweight autopilot with GPS, IMU, and pressure sensors is required. Manual override capability and wireless communication link are essential for AA241x as well as research projects. The same basic architecture satisfies these requirements.
Alternative: Single antenna GPS based attitude estimation techniques may be used for the GPS-only version.
Software Modules: Variometer
Sailplane pilots frequently use variometers to infer thermal updraft velocity. A variometer is an instrument that measures the total energy of an aircraft.
Characterizing gusts experienced by small UAVs: A comparison of the Power Spectral Density (PSD) of a measured gust and the Dryden PSD is shown below.
Acknowledgements
Dr. Hak-Tae Lee and Andre Marta provided valuable assistance with hardware selection and prototyping. Members of ADG helped significantly with fabrication and flight testing.
A similar functionality is built into the software for use in gust estimation and soaring related tasks. The rate of change of E closely follows the vertical gust velocity.
Chinmay Patel, Aircraft Aerodynamics and Design Group (ADG) Stanford University
Abstract
This poster provides an overview of the development of a low cost autopilot for use in research and coursework related to small autonomous Unmanned Aerial Vehicles (UAVs). Important specifications and software modules are discussed along with a control law development framework. Flight test results are shown. The capabilities of the autopilot make it an ideal choice for small UAVs and related projects.
April 24, 2007
Specifications
The specifications for the 2nd generation autopilot are shown below:
• 29.5 MHz processor • High level C-like programming language • GPS unit with 4 Hz update rate • Two-way wireless link with ~1 mile range • Inbuilt manual override capability • Inbuilt servo PWM generator • Small size: 2.5 x 2.75 x 1.0 inch • Weight: 2.3 oz including sensors, battery pack, and wiring • Sensors tailored for small UAVs • Airspeed, barometric altitude, 6-axis IMU
Control Law Design
Control laws are designed using a non-linear dynamics model in Simulink®. A PID control example is shown here. Sensor and actuator characteristics are also modeled where necessary.
Capabilities and Future Work
Autonomous pursuit flight: (Video) One airplane, always under manual control, broadcasts its GPS location every second. A second airplane tries to autonomously follow the first airplane. Results with a very simple position feedback based control strategy are encouraging. Some of the capabilities of the autopilot are listed below: • A wide range of autonomous UAV missions, e.g., Autonomous thermal and gust soaring • Data logging, system ID and control law validation for unconventional aircraft • Multiple-UAV missions: • Formation flight • Collaborative search • Cooperative soaring • Convoy following and target tracking • Collision avoidance
Introduction
A low cost and lightweight autopilot is needed for a graduate course, AA241x, offered by Prof. Ilan Kroo and Prof. Juan Alonso. In only one quarter, student teams design, build, and fly UAV to fulfill a prescribed mission, using a GPS module as the only sensor. The ability to fabricate and modify the autopilot in-house is also needed.
Software Modules: Attitude Estimation
Body axes accelerations and angular rates measured by an IMU are used to estimate the attitude of the aircraft. An Extended Kalman Filter (EKF) based approach, shown below, is used. The rates are first integrated in time to get a predicted attitude. Accelerometer readings are used as measurements to correct the predicted value.
Results
Trajectory from a GPS waypoint navigation flight is shown below. This trajectory can be plotted in near real-time and also be integrated with terrain mapping software.
Conclusions
• The autopilot provides a lightweight and low cost subsystem for UAV related research at ADG and serves as a tool for coursework. • Students get valuable hands-on experience with in-field testing of UAVs, validation of control laws, and familiarize themselves with embedded electronics. • Since the autopilot is designed and assembled in the laboratory, it can be easily modified for project-specific needs. • Wireless communication links open up the possibility for research in multi-UAV systems.
For research projects, a lightweight autopilot with GPS, IMU, and pressure sensors is required. Manual override capability and wireless communication link are essential for AA241x as well as research projects. The same basic architecture satisfies these requirements.
Alternative: Single antenna GPS based attitude estimation techniques may be used for the GPS-only version.
Software Modules: Variometer
Sailplane pilots frequently use variometers to infer thermal updraft velocity. A variometer is an instrument that measures the total energy of an aircraft.
Characterizing gusts experienced by small UAVs: A comparison of the Power Spectral Density (PSD) of a measured gust and the Dryden PSD is shown below.
Acknowledgements
Dr. Hak-Tae Lee and Andre Marta provided valuable assistance with hardware selection and prototyping. Members of ADG helped significantly with fabrication and flight testing.
A similar functionality is built into the software for use in gust estimation and soaring related tasks. The rate of change of E closely follows the vertical gust velocity.
