Autopilot Quadcopter

Khoa Phan-Dang
Hochiminh City University of Technology
Ho Chi Minh, Viet Nam
See full Bio Contact
Entry date: 12-Aug-2013
Final Submission: 01-Jul-2014
Quadcopter is a common platform for automatic control algorithm testing. In this project, we’re building a quadcopter platform on which several control algorithms would be experimented, then a camera and some other sensors would be added to complete an Autopilot model.

Project details


The system’s construction can be seen in figure 1. The Toradex computer module will be the center processing unit of the whole system, including image processing, communication management and autopilot execution. The Toradex module sends data from camera and commands from ground station to APM for auto-driving the quadcopter. The computer board will also read IMU and GPS data from the APM board to process and send it back to the ground station. The sonar sensor supplies information of altitude in order to maintain the quadcopter’s 3D-position in the air. 


Figure 1: System block diagram



The NVIDIA® Tegra™ based COM Kit with the NVIDIA's Tegra Graphics Processing Unit (GPU) would be a great advantage for image processing. Visual data helps keep the model’s position to follow check points or track lines, etc.

Figure 2 shows an typical Autopilot mode using image processing to follow check points.

 Figure 2: Autopilot typical mode using image processing


The testing platform consists of basic components of a quadcopter: the main frame, four DC brushless motors and ESCs (Electric Speed Controller), mainboard for controlling and signal processing (currently the APM 2.0).

Figure 3 shows a typical quadcopter structure 

Figure 3: Structure of a quadcopter



The PID control loop diagram currently used in the APM board is shown in Figure 3. Input signal RC_ROLL, RC_PITCH and RC_YAW from RF receiver or computer control board (embedded board) are updated in period of one milisecond and then compared to the IMU feedback data to drive and stabilize the model.


The model has been tested with manual control using the PID controller associated with the APM 2.0 and is now ready for control algorithm experiment as well as autopilot planning  (figure 5).

The next step is setting up the ground station using RF2500 module from Texas Instruments and MAVLink protocol.

Some control algorithms will be implemented and tested thanks to the data sent to ground station while in flight.

The final step is to put on a camera and using image processing to complete the Autopilot model

Figure 5: Current model


Updates Hide

Update #10035  |  29 Aug 2013
We're building a case for it right now, thank you Toradex

Comments Hide

Tran Truong Thanh Hai

good job team, seem profesional


Good job .. i also want to make an Autonomus Quad ..please help me !