Experimental Autonomous Vehicle Research Articles

Development and Testing of a Collision Avoidance Braking System for an Autonomous Vehicle

The article describes the main development and testing aspects of an emergency braking function for an autonomous vehicle. The purpose of this function is to prevent the vehicle from collisions with obstacles, either stationary or moving. An algorithm is proposed to calculate deceleration for the automated braking, which takes into account the distance to the obstacle and velocities of both the vehicle and the obstacle. In addition, the algorithm adapts to deviations from the required deceleration, which are inevitable in the real-world practice due to external and internal disturbances and unaccounted dynamics of the vehicle and its systems. The algorithm was implemented as a part of the vehicle’s mathematical model. Simulations were conducted, which allowed to verify algorithm’s operability and tentatively select the system parameters providing satisfactory braking performance of the vehicle. The braking function elaborated by means of modeling then was connected to the solenoid braking controller of the experimental autonomous vehicle using a real-time prototyping technology. In order to estimate operability and calibrate parameters of the function, outdoor experiments were conducted at a test track. A good consistency was observed between the test results and simulation results. The test results have proven correct operation of the emergency braking function, acceptable braking performance of the vehicle provided by this function, and its capability of preventing collisions.

Using Real World Data in Virtual Development and Testing of a Path Tracking Controller for an Autonomous Vehicle

The article describes an approach to development and testing of a path tracking function for an autonomous vehicle. The essence of the approach consists in combining experimental data and mathematical modeling in order to simulate operation of a path-tracking regulator in real world maneuvers. The procedure can be divided into two stages. The first one implies field-testing of the vehicle under control of a human driver with logging of the essential dynamic variables including the driving trajectory. Then the obtained data is used to validate the model of vehicle dynamics being a tool for further simulations. At the second stage, a simulation is performed with tracking of the previously logged trajectory by an automatic regulator. The results of these steps allow for comparison between the human and automatic controls with assessment of pros and cons of the latter and the ways of improving its performance. The proposed approach was implemented within a research and development project aimed at building of an experimental autonomous vehicle. The article describes the obtained results as well as the experiments and the mathematical model used for implementation of the said approach.

Logical artificial intelligence Mivar technologies for autonomous road vehicles

The main field of our research is artificial intelligence already implemented in the experimental autonomous road vehicle. Artificial intelligence now is a really huge area of studies. This investigation is focused more on a development of intelligent systems. Such as expert systems, semantic text processing systems, semantic image understanding systems, robotics systems. Wi!Mi is a solution for developing mivar knowledge models. This tool belongs to the class of so-called ES shells - programs that allow significantly simplify and accelerate the ES development process. It uses a novel knowledge representation mechanism based on bipartite graphs and the significantly improved but simple inference engine, which is able to process large systems involving millions of variables in just a few seconds. A special representation model which is called mivar nets with linear computational complexity using if-then rules is used in Wi!Mi. The idea is that the subject domain is divided into objects and connections between them. Scientific investigation allows to implement new decision-making system called “RoboRazum” as a form of an embedded software platform with flexible capabilities to adapt to control by any robotic suites and systems. RoboRazum is a platform for designing autonomous robotic control systems. Provided solutions don’t require powerful computational resources and could be used in small systems-on-chip like microcontrollers. That is why this solution is perfectly fit with conception of Internet of Things (IoT). Reasoned and fast logical decisions based on mivar technologies play important role for autonomous road vehicles.

Experimental Autonomous Road Vehicle with Logical Artificial Intelligence

This article describes some technical issues regarding the adaptation of a production car to a platform for the development and testing of autonomous driving technologies. A universal approach to performing the reverse engineering of electric power steering (EPS) for the purpose of external control is also presented. The primary objective of the related study was to solve the problem associated with the precise prediction of the dynamic trajectory of an autonomous vehicle. This was accomplished by deriving a new equation for determining the lateral tire forces and adjusting some of the vehicle parameters under road test conductions. A Mivar expert system was also integrated into the control system of the experimental autonomous vehicle. The expert system was made more flexible and effective for the present application by the introduction of hybrid artificial intelligence with logical reasoning. The innovation offers a solution to the major problem of liability in the event of an autonomous transport vehicle being involved in a collision.

A novel obstacle avoidance algorithm: “Follow the Gap Method”

In this paper, a novel obstacle avoidance method is designed and applied to an experimental autonomous ground vehicle system. The proposed method brings a new solution to the problem and has several advantages compared to previous methods. This novel algorithm is easy to tune and it takes into consideration the field of view and the nonholonomic constraints of the robot. Moreover the method does not have a local minimum problem and results in safer trajectories because of its inherent properties in the definition of the algorithm. The proposed algorithm is tested in simulations and after the observation of successful results, experimental tests are performed using static and dynamic obstacle scenarios. The experimental test platform is an autonomous ground vehicle with Ackermann steering geometry which brings nonholonomic constraints to the vehicle. Experimental results show that the task of obstacle avoidance can be achieved using the algorithm on the autonomous vehicle platform. The algorithm is very promising for application in mobile and industrial robotics where obstacle avoidance is a feature of the robotic system.

Experimental Autonomous Veidcle Systems - Requirements and a Prototype Architecture

The paper describes the requirements for and a prototype configuration of a software architecture for control of an experimental autonomous vehicle. The test bed nature of the system is emphasised in the choice of architecture making re-configurability, data logging and extendability simple. The central element of the architecture is the 'global database' that serves several purposes, such as storing system parameters, making signals available for data logging and inter-process communication. Standard software components are used to a large extent, 05-9 as real-time operating system, a custom ANSI-C program extending the TCL system is used for plan execution and a combination of MATLAB and a custom made Java GUI as user interface on the remote operator console. The choice of these standard software components is explained and the individual components demonstrated. Examples of how specific tasks are implemented using the architecture are given.

An Autonomous Running Vehicle Equipped with a Vision System for Fixing and Recognizing Color Landmarks.

An experimental autonomous running vehicle with a vision system is developed for load carriers in factories or activity support vehicles of aged or visually impaired people. The vision system extracts color objects of red, yellow, green, or blue in a scene and measures in video-rate the distance and the apparent area of them. By the vision system, the vehicle can recognize color landmarks and obeys the commands assigned to them as well as keeping away from obstacles using four supersonic range meters arranged around the vehicle. In addition, a fixation behavior to color landmarks is realized by a color CCD camera and two stepping mortors to move the camera in horizontal and vertical directions. The running control system is configurated by the Sub-Sumption Architecture which enables us to develop a complicated control system by combining control systems of different behavior. The applicability of the vehicle is demonstrated by several autonomous running experiments to run on the routes specified by color landmarks.

Autonomous Running Technique by Visual Servoing Considering Route Transition and Running Behavior.

Visual servoing is one type of vision-based robot movement control. Measurement of robot's own position is not required because the movement of the robot is controlled to match the image obtained by the vision sensors on the robot to a given image (teaching image) obtained beforehand. In this paper we propose an autonomous running technique of visual servoing in consideration of route transitionand rough running behavior to reach the position of obtaining a teaching image. The route transition technique is needed to avoid obstacles on the running route and contributes to reducing the data size of the route information abstracted from teaching images because the data of overlapping parts of teaching routes can be omitted. The sampling interval of teaching images can be made longer by considering the running behavior. This contributes to teaching data reduction. An experimental autonomous running vehicle is developed and the proposed technique is demonstrated.

An Autonomous Mobile Vehicle

This paper describes an experimental autonomous mobile vehicle which can move through an initially known obstacle field to a nominated goal. The navigation system uses combination of laser rangefinder and odometry for environmental mapping and localization, a GVD technique path planning algorithm and a pilot algorithm for motion control. The architecture of control system of the vehicle was designed as multilevel hierarchical distributed system which is built up as a modular structure. The philosophy of the hardware and software solution of individual modules of the control system and a description of the current state of the research is presented.

The steering control of an experimental autonomous vehicle

Experimental results relating to the steering control system of an experimental autonomous vehicle operating in a pre-defined and structured environment are presented. A method of steering control is formulated which relates second-order criteria, such as rise time and damping factor, to the parameters of the control system. With the control system described, the vehicle can track a path of constant heading with very little steady-state error and, if the control parameters are chosen in the correct ratio, it will respond in a critically damped fashion to the step disturbances which are generated by position corrections. For a curved path, demand curvature is treated as an input to the system in order to eliminate steady-state offset errors in tracking. In order to make the vehicle's performance independent of velocity, control parameters are derived as simple functions of velocity. This has the effect of allowing the vehicle to manoeuvre more tightly at lower speeds, and makes use of the increased availability of drive-system torque at such speeds. In order to prevent stalling of the stepper-motor drive system at large changes in demand curvature, a method of limiting the rate of steering is described.

Using Common LISP in the EAVE autonomous underwater vehicle

The Marine Systems Engineering Laboratory of the University of New Hampshire has ported the University of Utah's Portable Common LISP Subset (PLCS) to the experimental autonomous vehicle (EAVE) underwater autonomous vehicle. Using Common LISP in the EAVE autonomous vehicle is expected to improve programmer productivity and software portability. Using the LISP interpreter allows for software changes to be made while in the field, thus saving time during vehicle operations. Issues concerning the operation of LISP in a real-time environment, such as the impact of garbage collection (GC), have been resolved by using an efficient version of Common LISP and by using LISP) at the high-end of a time-based software hierarchy. >

Terrain data base generation for autonomous land vehicle navigation

The Defense Advanced Research Projects Agency (DARPA) Strategic Computing Program is a large, multi-year effort focused on developing the next generation of computers and machine intelligence. Within this program the Autonomous Land Vehicle (ALV) Project calls for the development and demostration of increasingly sophisticated autonomous land navigation capabilities. As part of this effort, the U.S. Army Engineer Topographic Laboratories (USAETL) has the task of producing a high-resolution, high-accuracy, experimental digital terrain data base of a 12-km 2 test site. This data base will be used in conjunction with inertial navigation and computer vision systems for autonomous traversal by the vehicle. The data base consist of six overlays including landforms, soils, surface drainage, land cover, roads, and a high-density digital elevation model (DEM). ALV application of the terrain data base include route planning and a-priori information for machine vision. Data compilation was performed on the Computer-Assisted Photo Interpretation Research (CAPIR) system at USAETL. CAPIR is an ongoing research effort which addresses the issues of digital terrain data extraction, storage, and exploitation. This paper willaddress data requirements and systems for building terrain dta bases in support of experimental autonomous land vehicle navigation.

Development of an Experimental Autonomous Land Vehicle

Development of an Experimental Autonomous Land Vehicle

Guidance and control architecture for the EAVE vehicle

For the past several years the Marine Systems Engineering Laboratory (MSEL) has directed its efforts towards the development of the technologies required for unmanned untethered submersible vehicles. The current focus of those efforts is to develop a system architecture that will allow the implementation of a knowledge-based guidance and control system. The goal of this effort is to implement a simple system which has addressed the basic problems and will allow for expansion as insight is gained from field testing the concepts using the Experimental Autonomous Vehicle (EAVE) system at MSEL. This paper considers those factors that have driven the development of an architecture which is being implemented in the EAVE vehicle system. Its intent is to focus on those issues that have guided the application of artificial intelligence (AI) techniques to meet the requirements of the system and its mission. The architecture being implemented is outlined and some of its features detailed.