Autonomous flight control software

Download Free PDF. Andon Topalov. A short summary of this paper. Download Download PDF. Translate PDF. Shakev, Sevil A. Ahmed, Andon V. Topalov, Vasil L. Popov and Kostadin B. Shiev 1 Introduction Unmanned aerial vehicles UAVs are becoming increasingly popular for different commercial and noncommercial applications in the areas of industry, entertainment, research, and military.

Recent technological achievements have led to their further miniaturization and increased capability for carrying on number of sensors while the duration of the flight can now last longer several decades of minutes or even hours.

On the other hand, Cyber Physical Systems CPS are getting more and more popular and powerful tool for implementing advanced solutions of old problems, relying on technologies such as embedded computing, Internet of Things IoT , Internet of Service IoS , cloud computing, big data, etc. The number of applications of CPS is growing in last few years. CPSs are successfully integrated with smart multi agent robotized systems for environmental surveillance, rescue robots, etc. The role of these systems in such missions is extremely important because they can provide N.

Ahmed e-mail: sevil. Topalov e-mail: topalov tu-plovdiv. Shiev e-mail: k. Sgurev et al. Shakev et al. The algorithms proposed in this work are inspired by the concept of CPSs. Using gestures for precise positioning of small UAVs is an innovative way of making the human part of real-time working CPSs.

Most of the UAVs, currently available on the market, have been designed to work in outdoor environments, relying on GPS data to determine their current position. Although well-developed GPS systems and services, momentary loose of coverage and missing data could became a serious problem. Moreover, there are many practical tasks where UAVs will be required to operate in indoor environments or along narrow city streets where GPS data cannot be received or are unreliable.

The inspection of buildings and equipment with dam- aged structures and the decommissioning of nuclear power plants are among them. Since such monitoring tasks can be difficult or dangerous for the engaged personnel, a distant inspection is frequently considered. Most of the currently existing robots capable to perform such activities are ground based vehicles and can inspect only the lower zones in these environments [1].

Furthermore, inspections could be difficult and time consuming process that may include several stages like characterization, decontamination, demolition and waste management. It requires collection of sig- nificant amount of information about the ruggedness and structural integrity of the building or the monitored equipment [2].

There might be a necessity to determine also the quantity, disposition and intensity of the emissions of radioactive materials or dangerous chemicals. An UAV capable to perform such inspection tasks must maintain stable flight, avoid obstacles, localize position and orientation and plan its way without a GPS signal.

Human operator must have a possibility to send instructions for additional maneuvers in order to obtain better information about the inspected items. Several basic problems have to be solved at the design stage of such aerial robot: i construc- tion of suitable vehicle platform, capable of carrying the required payload; ii UAV stabilization and control of the movements during the flight; iii calculation of the current position and orientation; iv planning the path in order to reach a specified target position.

The first problem could be resolved through the purchase of an available on the market aerial platform. A number of well-designed quadrotor rotorcrafts, ranging from the professional ones [3], to those intended for hobbyists [4] are currently available. Many of them are stabilized and have good control on the dynamics of the flight.

In spite of that, for environments where the GPS signal is absent or unreliable, the position and orientation determination issues as well as the path planning are currently an active area of research. Autonomous Flight Control and Precise … Gestural control is getting more and more popular with introduction of collabora- tive features of modern robots. In last 5 years many researchers devote big attention to gestures defining and recognition with idea to use them in different robot con- trol strategies—beginning from basic movements to fine effector control [9—13].

This type of remote control applied by an operator is stated as a variety of natural user interface NUI used in man-machine communication [14, 15]. The idea is not a new one.

In particular, a group of researchers from the Institute for Dynamics and Control at ETH Zurich has proposed a way to dynamically interact with quadrotors based on the hands posi- tion of the operator [16]. The approach adopted here differs from [16] in this that the recognized gestures are used to launch discrete control commands to the quadrotor and absence of external tracking system to determine its position and orientation. Drone quadrotor after switching it in semi-autonomous mode.

The human-rotorcraft interaction is done through gestures and body postures recognized by Microsoft Kinect. Drone quadrotor is a commercially available electrically powered four-rotor mini-rotorcraft with dimensions of mm by mm and weight of g, initially designed for playing augmented reality games [17]. The quadrotor consists of a carbon-fiber support structure, four high-efficiency brushless motors, sensor and control board, two cameras, and is equipped with indoor and outdoor removable hulls.

The power supply is provided by a 3-cell, mAh lithium- polymer battery that suffices for approximately 15 min long flights. The rotorcraft has a set of onboard sensors including ultrasonic altimeters measuring the distance to the land surface, and thus helping to maintain stability when hovering at up to 6 m, a 3-axis accelerometer together with a 2-axis gyro used to provide accurate pitch and roll angles, and 1-axis high precision gyro to measure the yaw angle.

The AR. The communication with the drone can be done via an ad hoc WiFi network using the provided software interface. An external computer can connect to AR. The communication can be done via three channels, each with a different UDP port. Over the command channel the drone, can be requested to take off and land, to change configuration of its controllers, to calibrate sensors etc.

The navdata channel sends information about the status of the quadrotor and pre-processed sensory data con- taining current yaw, pitch and roll angles, the altitude, battery state and 3D speed estimates. Both channels update the transmitted data at 30 Hz rate. The stream chan- nel provides images from the frontal and bottom cameras. The control board of the AR. Drone is accessible by telnet and changes can be made in the settings of the on-board operating system and in the configuration files of the internal controllers.

The lifting force is generated by four propellers. Unlike most rotorcrafts, quadro- tors use fixed pitched propellers. The angular velocity variations between the propellers cause lift and torque thus motion.

Autonomous Flight Control and Precise … Fig. Drone quadrotor there are built-in onboard control loops for the attitude and vertical speed stabilization. The existence of the build-in controllers imposes a requirement to modify the control problem since it becomes necessary to model the dynamics of the entire aerial vehicle together with the built-in control loops system.

In this case the trajectory tracking control problem can be defined as to find the appropriate reference values of the vertical speed, yaw, pitch, and roll angles that will allow the rotorcraft to follow desired positions in the space see Fig. Since u is not directly accessible it cannot be used as a control signal. The block diagram is shown on Fig. To achieve a stable closed loop dynamic system with acceptable dynamic prop- erties the gains of the controller have to be tuned appropriately.

The selected pulse width was 40 ms, and the pulse period was 10 s. The graphical results are represented on Fig. The block diagram showing the control loop for the lateral movements is presented on Fig. The parameters of the external PD controller for lateral movements are similar to those obtained for forward movements controller as shown on Table 1. Since the control task is very com- plex the operator has to control six degrees of freedom using four control signals , it was found that the system achieves better performance if the control signals are with fixed values.

Thus the semi-autonomous control system has to receive from the gestural inter- face system only the movement directions, to execute the desired movements with predefined constant velocities while keeping the rotorcraft attitude. The structure of the developed gestural control strategy is presented on Fig. The C program is configured as an asynchronous socket server and sends gesture data to the Simulink model, which controls the drone by passing the recognized gestures.

They read the skele- ton information over the described above data streams and recognize the gestures according to the active joints of the skeleton and the rules from Tables 2, 3 and 4. The recognized gestures and body postures are then used for the remote control of the AR. Drone quadrotor. The information from Kinect is structured in three basic data streams: depth, color, and audio. In this post, we will be looking at some of the best open source uav projects out there.

After reading this, you will understand which drone firmware projects are viable for the drone you are wanting to create. Have I successfully totally confused Google regarding what the heck this page is about?? Ardupilot was one of the early pioneers of open source drone software. The ArduPilot community is really supportive, and their forum is a great tool for getting involved. ArduPilot can work on many different types of vehicles, including: multirotors, fixed wing planes, land rovers and even submarines.

Ardupilot strives to make these vehicles purely autonomous. There is also a diverse collection of flight controller boards that ArduPilot is compatible with. The open source license that ArduPilot operates by is GPL, which essentially means that any changes that are made to the ArduPilot codebase needs to be added back to the parent project.

Then they could keep their innovation private, while using ArduPilot for the basic function of drone control. While you can use ArduPilot for some FPV quad applications, typically it is used by those wanting a drone that can fly autonomously. I have written an in-depth overview of the ArduPilot flight stack for those interested. PX4 is a part of the Dronecode project, a non-profit organization administered by the Linux Foundation.

Dronecode aims to supply the emerging drone industry with a platform of open source software. An interesting anecdote is that ArduPilot used to be the featured flight control software project in Dronecode, which has contributed to an ArduPilot vs PX4 semi-rivalry. PX4 came out of the Pixhawk project, which is an open source hardware project, as they needed some open source drone software to run their boards. This feature makes PX4 more attractive to businesses looking to protect their intellectual property.

Due to this reason, and their inclusion in the Dronecode project, PX4 is getting much more funding for development work than ArduPilot. PX4 is also supported on a large amount of flight control boards. The BetaFlight flight controller software is focused on the performance of manual flight, making it a great choice for FPV quad fanatics. It is arguably the most popular open source software project for FPV drones today. BefaFlight is primarily used with quadcopters, but can be used on fixed wing aircraft as well.