Research efforts on the development of autonomous vehicles date back to the 1920s and recent announcements indicate that those cars are close to becoming commercially available. However, the most successful prototypes that are currently being demonstrated rely on an expensive set of sensors. This study investigates the use of an affordable vision system as a planner for the Robocart, an autonomous golf cart prototype developed by the Wireless Innovation Laboratory at WPI. The proposed approach relies on a stereo vision system composed of a pair of Raspberry Pi computers, each one equipped with a Camera Module. They are connected to a server and their clocks are synchronized using the Precision Time Protocol (PTP). The server uses timestamps to obtain a pair of simultaneously captured images. Images are processed to generate a disparity map using stereo matching and points in this map are reprojected to the 3D world as a point cloud. Then, an occupancy grid is built and used as input for an A* graph search that finds a collision-free path for the robot. Due to the non-holonomic constraints of a car-like robot, a Pure Pursuit algorithm is used as the control method to guide the robot along the computed path. The cameras are also used by a Visual Odometry algorithm that tracks points on a sequence of images to estimate the position and orientation of the vehicle. The algorithms were implemented using the C++ language and the open source library OpenCV. Tests in a controlled environment show promising results and the interfaces between the server and the Robocart have been defined, so that the proposed method can be used on the golf cart as soon as the mechanical systems are fully functional.

Quadrotors, being more agile than fixed-wing vehicles, are the ideal candidates for autonomous missions in small, compact spaces. The immense challenge to navigate such environments is fulfilled by the concept of path following. Path following is the method of tracking/tracing a fixed, pre-defined path with minimum position error while exerting the lowest possible control effort. In this work, the missile guidance technique of pure pursuit is adopted and modified for a 3D quadrotor model to follow fixed, compact trajectories. A specialized hardware testing platform is developed to test this algorithm. The results obtained from simulation and flight tests are compared to results from another technique called differential flatness. A small part of this thesis also deals with the stability analysis of the modified 3D pure pursuit algorithm to track trajectories expending lower control effort.

The automotive industry has been involved in making vehicles autonomous to different levels in the past decade rapidly. Particularly in the commercial vehicle market, there is a significant necessity to make trucks have a certain level of automation to help reduce dependence on human efforts to drive. This could help in reducing several accidents caused by human error. Interestingly there are several challenges and solutions in achieving and implementing autonomous driving for trucks. First, a benchmark of different control architectures that can make a truck drive autonomously are explored. The chosen controllers (Pure Pursuit, Stanley, Linear Quadratic Regulator, Sliding Mode Control and Model Predictive Control) vary in their simplicity in implementation and versatility in handling different vehicle parameters and constraints. A thorough comparison of these path tracking controllers are performed using several metrics. Second, a collision avoidance system based on cubic polynomials, inspired by rapidly exploring random tree (RRT) is presented. Some of the path tracking controllers are limited by their ability and hence a standalone collision avoidance system is needed to provide safe maneuvering. Simulations are performed for different test cases with and without obstacles. These simulations help compare safety margin and driving comfort of each path tracking controller that are integrated with the collision avoidance system. Third, different performance metrics like change in acceleration input, change in steering input, error in path tracking, deviation from base frame of track file and lateral and longitudinal margin between ego and target vehicle are presented. To conclude, a set of suitable controllers for heavy articulated vehicles are developed and benchmarked. / Bilindustrin har varit involverad i att göra fordon autonoma till olika nivåer under det senaste decenniet snabbt. Särskilt på marknaden för kommersiella fordon finns det ett stort behov av att få lastbilar att ha en viss nivå av automatisering för att minska beroendet av mänskliga ansträngningar att köra. Detta kan hjälpa till att minska flera olyckor orsakade av mänskliga fel. Intressant nog finns det flera utmaningar och lösningar för att uppnå och implementera autonom körning för lastbilar. Först utforskas ett riktmärke av olika styrarkitekturer som kan få en lastbil att köra autonomt. De valda kontrollerna (Pure Pursuit, Stanley, Linear Quadratic Regulator, Sliding Mode Control och Model Predictive Control) varierar i sin enkelhet i implementering och mångsidighet när det gäller att hantera olika fordonsparametrar och begränsningar. En grundlig jämförelse av dessa vägspårningskontroller utförs med hjälp av flera mätvärden. För det andra presenteras ett system för undvikande av kollisioner baserat på kubiska polynom, inspirerat av snabbt utforskande slumpmässiga träd (RRT). Vissa av vägspårningskontrollerna är begränsade av sin förmåga och därför behövs ett fristående system för att undvika kollisioner för att ge säker manövrering. Simuleringar utförs för olika testfall med och utan hinder. Dessa simuleringar hjälper till att jämföra säkerhetsmarginal och körkomfort för varje vägspårningskontroller som är integrerade med kollisionsundvikande systemet. För det tredje presenteras olika prestandamått som förändring i accelerationsinmatning, förändring i styrinmatning, fel i banspårning, avvikelse från basramen för spårfilen och lateral och longitudinell marginal mellan ego och målfordon. Avslutningsvis utvecklas och benchmarkas en uppsättning lämpliga styrenheter för tunga ledade fordon.

This thesis work contains a stabilization and a motion planning strategy for a truck and trailer system. A dynamical model for a general 2-trailer with two rigid free joints and a kingpin hitching has been derived based on previous work. The model holds under the assumption of rolling without slipping of the wheels and has been used for control design and as a steering function in a probabilistic motion planning algorithm. A gain scheduled Linear Quadratic (LQ) controller with a Pure pursuit path following algorithm has been designed to stabilize the system around a given reference path. The LQ controller is only used in backward motion and the Pure pursuit controller is split into two parts which are chosen depending on the direction of motion. A motion planning algorithm called Closed-Loop Rapidly-exploring Random Tree (CL-RRT) has then been used to plan suitable reference paths for the system from an initial state configuration to a desired goal configuration with obstacle-imposed constraints. The motion planning algorithm solves a non-convex optimal control problem by randomly exploring the input space to the closed-loop system by performing forward simulations of the closed-loop system. Evaluations of performance is partly done in simulations and partly on a Lego platform consisting of a small-scale system. The controllers have been used on the Lego platform with successful results. When the reference path is chosen as a smooth function the closed-loop system is able to follow the desired path in forward and backward motion with a small control error. In the work, it is shown how the CL-RRT algorithm is able to plan non-trivial maneuvers in simulations by combining forward and backward motion. Beyond simulations, the algorithm has also been used for planning in open-loop for the Lego platform. / <p>Links to movies:</p><p>Reference tracking on Lego platform:</p><p>https://www.dropbox.com/s/ebtfgfo7aq9ij8w/reference_tracking.m4v?dl=0</p><p></p><p>Motion planning simulation with CL-RRT:</p><p>https://www.dropbox.com/s/z9kk27cxdxc1llp/CL_RRT_motion_planning.wmv?dl=0</p>

The aim of this project was to compare two control algorithms designed to steer an autonomous vehicle. The comparison was made using a simulated environment to evaluate the performance of both controllers. The simulation used in this project was designed in Python and used an algorithm which randomly constructed roads from predefined road segments to create paths for the vehicle to follow. In this environment the Linear Time Varying (LTV)-Model Predictive Controller (MPC) and Pure Pursuit Controller (PPC) algorithms were evaluated. The thesis compared how well they follow paths, the average control cost of completing tasks, how well they handle input constraints, and the computational time for each algorithm. The data was collected by driving along three sets of randomly generated roads with both control algorithms. One set mostly straight, one with some turns, and one with mostly turns. An Analysis of Variance (ANOVA) test was used to make the comparison between the performance of the two algorithms. The results showed that both algorithms performed well. The PPC had low computation time and used less control, but it also had larger position errors. The LTV-MPC had higher computation time, but smaller position errors at the cost of larger control values. The conclusion is that the MPC is preferable if computational capabilities are available. Room for future work exists in the form of comparing additional controller types for autonomous vehicles and exploring different tuning parameters for the MPC controller. The simulation could also be expanded to more accurately reflect real world conditions. / Målet med detta projekt var att jämföra två kontrollalgoritmer avsedda för att styra en självkörande bil. Jämförelsen gjordes med hjälp av en simulering som utformades i Python. Den använde sig av en algoritm som slumpmässigt satte ihop vägar från förkonstruerade delar för att skapa banor för den självkörande bilen att följa. I denna miljö har vi testat två algoritmer, en LTV-MPC och en PPC. Vi jämförde hur pass väl de följer banor som skall likna riktiga vägar, hur mycket styrning de använder sig av för att bedöma energianvändning, hur väl de förhåller sig till begränsningar på acceleration och styrning, och den beräkningstiden som krävdes för att köra vår algoritm. Datan samlades genom att köra längs med tre grupper av slumpmässigt genererade vägar med båda kontrollalgoritmerna. En grupp innehöll huvudsakligen raka sträckor, en innehöll en del svängar, och en innehöll mycket svängar. ANOVA-testet användes för att göra jämförelsen mellan resultatet av dessa två algoritmer. Resultatet visade att båda algoritmer presterar väl. PPCn hade låg beräkningstid och mindre styrvärden, men större positionsfel. MPCn hade högre beräkningstid och större styrvärden, men mindre positionsfel. Slutsatsen är att MPCn är att föredra om beräkningsmöjligheterna finns tillgängliga. Det finns utrymme för framtida arbete i form av att jämföra fler kontrollalgoritmer och att utforska fler parameter justeringar för MPCn. Utöver det finns det även utrymme för en simulation som reflekterar verkligheten noggrannare.

No description available.