Before vehicles that go to sale for public it goes through many stages of testing in different fields. All types of pre-production cars are the vehicles that come after prototypes which allow the OEM to find out possible issues by running different categories of tests. One of the most important fields of test is regarding safety. When it comes to autonomous cars, where safety is crucial, the number of tests to carry out, time for handling them and their complexity is increased. To reduce costs, complexity and time some of them might be simulated. After the analysis is done the cars have to be tested in real time to collect more data, not only from car sensors but also through observation from the tests filmed. This thesis contains an overview of whole project leading to develop an autonomous platform, driving after autonomous car and controlled by simulation environment and solving major issues of the project. The filming of autonomous cars remotely had two major issues. One of the issues was to integrate all the different software and platforms. The trajectories of the autonomous cars were controlled in different software and the trajectories of the RC were managed in different software. The solution to integrate all the software and platforms is shown in this thesis. The second issue was to make the RC car to follow the waypoints that are generated by simulation and by communicating with real car using a PID controller. The project is prepared for future improvements, like installing a camera to the RC car which will follow predefined test independently.

The industry of UAVs has experienced a boost in recent years, and developments on both the hardware and algorithmic side have enabled smaller and more accessible drones with increased functionality. This thesis investigates the possibilities of autonomous target recognition and tracking in small, low-cost drones that are commercially available today. The design and deployment of an object recognition and tracking algorithm on a Crazyflie 2.1, a palm-sized quadcopter with a weight of a few tens of grams, is presented. The hardware is extended with an expansion board called the AI-deck featuring a fixed, front-facing camera and a GAP8 processor for machine learning inference. The aim is to create a vision-based autonomous control system for target recognition and following, with all computations being executed onboard and without any dependence on external input. A MobileNet-SSD object detector trained for detecting human bodies is used for detecting a person in images from the onboard camera. Proportional controllers are implemented for motion control of the Crazyflie, that process the output from the detection algorithm to move the drone to the desired position. The final implementation is tested indoors and proved to be able to detect a target and follow simple movements of a human moving in front of the drone. However, the reliability and speed of the detection need to be improved to achieve a satisfactory result.

Thesis (MScEng (Electrical and Electronic Engineering))--University of Stellenbosch, 2010. / ENGLISH ABSTRACT: This thesis presents the development and implementation of a position based kinematic guidance system, the derivation and testing of a Dynamic Pursuit Navigation algorithm and a thorough analysis of an aircraft’s runway interactions, which is used to implement automated take-off of a fixed wing UAV. The analysis of the runway is focussed on the aircraft’s lateral modes. Undercarriage and aerodynamic effects are first analysed individually, after which the combined system is analysed. The various types of feedback control are investigated and the best solution suggested. Supporting controllers are designed and combined to successfully implement autonomous take-off, with acceleration based guidance. A computationally efficient position based kinematic guidance architecture is designed and implemented that allows a large percentage of the flight envelope to be utilised. An airspeed controller that allows for aggressive flight is designed and implemented by applying Feedback Linearisation techniques. A Dynamic Pursuit Navigation algorithm is derived that allows following of a moving ground based object at a constant distance (radius). This algorithm is implemented and verified through non-linear simulation. / AFRIKAANSE OPSOMMING: Hierdie tesis handel oor die ontwikkeling en toepassing van posisie-afhanklike, kinematiese leidings-algoritmes, die ontwikkeling van ’n Dinamiese Volgings-navigasie-algoritme en ’n deeglike analise van die interaksie van ’n lugraam met ’n aanloopbaan sodat outonome opstygprosedure van ’n vastevlerk vliegtuig bewerkstellig kan word. Die bogenoemde analise het gefokus op die laterale modus van ’n vastevlerk vliegtuig en is tweeledig behartig. Die eerste gedeelte het gefokus op die analise van die onderstel, terwyl die lugraam en die aerodinamiese effekte in die tweede gedeelte ondersoek is. Verskillende tipes terugvoerbeheer vir die outonome opstygprosedure is ondersoek om die mees geskikte tegniek te bepaal. Addisionele beheerders, wat deur die versnellingsbeheer gebaseerde opstygprosedure benodig word, is ontwerp. ’n Posisie gebaseerde kinematiese leidingsbeheerstruktuur om ’n groot persentasie van die vlugvermoë te benut, is ontwikkel. Terugvoer linearisering is toegepas om ’n lugspoedbeheerder , wat in staat is tot aggressiewe vlug, te ontwerp. ’n Dinamiese Volgingsnavigasie-algoritme wat in staat is om ’n bewegende grondvoorwerp te volg, is ontwikkel. Hierdie algoritme is geïmplementeer en bevestig deur nie-lineêre simulasie.