External drive-by-wire kit and control architecture for autonomous and teleoperated vehicles

As of today, autonomous driving is an area of research, it is in continuous evolution and most importantly it is not yet completely regulated and standardized. That is, full self-driving cars still cannot circulate without a human supervisor, with some exceptions. Instead of worrying about the full autonomous drive stack, this thesis work aims to enter in such area in a different way by realizing a mountable kit enabling drive-by-wire features. The goal is to create a modular, low-level, hardware and software interface for autonomous and teleoperated driving implementations that can be mounted as an external kit on any vehicle without any substantial or critical alteration, even on commercial ones. In this way there would be no need neither to buy a prepared car nor to enter in the CAN of the vehicle which, for obvious reason, is proprietary and not developed to be accessed from the outside. The kit has been designed as a set of devices, that is this system wants to be modular and scalable by adding or removing specific modules in order to adapted as needed. The kit manages everything from the low-level stuff, that are the control loops and the real-time operations, to the path follower and communication with a possible remote base station. Higher level operations don’t concern the study. The thesis is therefore focused on such system architecture design and implementation, the definition of the levels and inner interfaces, reaching the exposure of the high-level one to the system user, and on the communication and teleoperation module, allowing the users to care only about the mission planning and its management. In this way, a possible user can utilize this system by connecting to it with, for example, a proprietary interface for teleoperating instead of the given one as well as through a perception level and its path planner. The study, at the moment, has not been done with the goal to work in urban and dense environments but in partially unknown ones, outdoor or not populated areas, in situations where humans can be in danger like, for example, rescue scenarios or humanitarian missions in dangerous areas. In such case a teleoperated or autonomous vehicle would be sent instead, revealing its usefulness when safety of operators is needed. This kit aims to be a versatile and cost-effective system with respect to an adapted vehicle that would guarantee such safety, since its main components are applicable and adjustable to all vehicles in order to make economy of scale and reduce costs. Nonetheless the project lends itself well to be used and further developed also for other scenarios. A study on the possible communication channels has also been conducted with the idea to create a completely independent and self-sufficient transmission system, which was then revealed, for now, to be infeasible due to latency and bandwidth limits. For the same reasons, the teleoperation interface and its functionality is designed with a minimalist approach in a way to work even with low bandwidth available. The vision and perception part that is used is not a subject of this thesis. The system has been tested and it is operative. This project aims to be a starting point and future works will then be considered.