Arda Kurt

Cooperative Adaptive Cruise Control Implementation of Team Mekar at the Grand Cooperative Driving Challenge

This paper presents the cooperative adaptive cruise control implementation of Team Mekar at the Grand Cooperative Driving Challenge (GCDC). The Team Mekar vehicle used a dSpace microautobox for access to the vehicle controller area network bus and for control of the autonomous throttle intervention and the electric-motor-operated brake pedal. The vehicle was equipped with real-time kinematic Global Positioning System (RTK GPS) and an IEEE 802.11p modem installed in an onboard computer for vehicle-to-vehicle (V2V) communication. The Team Mekar vehicle did not have an original-equipment-manufacturer-supplied adaptive cruise control (ACC). ACC/Cooperative adaptive cruise control (CACC) based on V2V-communicated GPS position/velocity and preceding vehicle acceleration feedforward were implemented in the Team Mekar vehicle. This paper presents experimental and simulation results of the Team Mekar CACC implementation, along with a discussion of the problems encountered during the GCDC cooperative mobility runs.

Hybrid-state driver/vehicle modelling, estimation and prediction

The first part of this study develops a general architecture for estimation and prediction of hybrid-state systems. The proposed system utilizes the hybrid characteristics of decision-behaviour coupling of many systems such as the driver and the vehicle; uses estimates of observable parameters to track instantaneous discrete state and predicts the most likely outcome, depending on the discrete model and the observed behaviour of the continuous subsystem. The proposed method is suitable for the scenarios that involve unknown decisions of other individuals, such as lane changes or intersection precedence/access. In the second part, this paper specifically deals with the implementation of the proposed methodology on an intersection safety system, predicting the vehicle behaviours and potential outcomes through traffic intersection scenarios. Driver intentions are tracked and predicted through vehicle behaviour, and possible combinations of intention predictions for different vehicles are interpreted for the safety of the situation.

Hybrid State System Development for Autonomous Vehicle Control in Urban Scenarios

Abstract This paper analyzes a hybrid-state-system-based controller for an autonomous vehicle in urban traffic and provides development procedures for hybrid-state systems for automatic control. The Ohio-State University Autonomous City Transport utilizes a discrete-state system, based on a finite state machine for high-level decision making and a continuous-state controller for low-level lateral and longitudinal control. The design procedure for the overall hybrid controller involves a series of capability grafts, each improving the ability of the vehicle to handle diverse situations. The design methodology, as demonstrated in a number of development steps, and architecture are capable of handling various urban scenarios, as demonstrated in a June 2007 site visit by Darpa officials.

A sampling-based local trajectory planner for autonomous driving along a reference path

In this paper, a state space sampling-based local trajectory generation framework for autonomous vehicles driving along a reference path is proposed. The presented framework employs a two-step motion planning architecture. In the first step, a Support Vector Machine based approach is developed to refine the reference path through maximizing the lateral distance to boundaries of the constructed corridor while ensuring curvature-continuity. In the second step, a set of terminal states are sampled aligned with the refined reference path. Then, to satisfy system constraints, a model predictive path generation method is utilized to generate multiple path candidates, which connect the current vehicle state with the sampling terminal states. Simultaneously the velocity profiles are assigned to guarantee safe and comfort driving motions. Finally, an optimal trajectory is selected based on a specified objective function via a discrete optimization scheme. The simulation results demonstrate the planners capability to generate dynamically-feasible trajectories in real time and enable the vehicle to drive safely and smoothly along a rough reference path while avoiding static obstacles.

Lane Change and Merge Maneuvers for Connected and Automated Vehicles: A Survey

Intelligence in vehicles has developed through the years as self-driving expectations and capabilities have increased. To date, the majority of the literature has focused on longitudinal control topics (e.g. Adaptive Cruise Control (ACC), Cooperative ACC (CACC), etc.). To a lesser extent, there have been a variety of research articles specifically dealing with lateral control, e.g., maneuvers such as lane changes and merging. This paper provides a survey of this particular area of vehicle automation. The key topics addressed are control systems, positioning systems, communication systems, simulation modeling, field tests, surroundings vehicles, and human factors. Overall, there has been some successful research and field testing in lane change and merge maneuvers; however, there is a strong need for standardization and even more research to enable comprehensive field testing of these lateral maneuvers, so that commercial implementation of automated vehicles can be realized.

Driver/vehicle state estimation and detection

The authors present a cyber-physical systems related study on the estimation and prediction of driver states in autonomous vehicles. The first part of this study extends on a previously developed general architecture for estimation and prediction of hybrid-state systems. The extended system utilizes the hybrid characteristics of decision-behavior coupling of many systems such as the driver and the vehicle; uses Kalman Filter estimates of observable parameters to track the instantaneous discrete state, and predicts the most likely outcome. Prediction of the likely driver state outcome depends on the higher level discrete model and the observed behavior of the continuous subsystem. Two approaches to estimate the discrete driver state from filtered continuous observations are presented: rule based estimation, and Hidden Markov Model (HMM) based estimation. Extensions to a prediction application is described through the use of Hierarchical Hidden Markov Models (HHMMs). The proposed method is suitable for scenarios that involve unknown decisions of other individuals, such as lane changes or intersection precedence/access. An HMM implementation for multiple tasks of a single vehicle at an intersection is presented along with preliminary results.

Driver behavior modeling near intersections using support vector machines based on statistical feature extraction

The capability to estimate drivers intention leads to the development of advanced driver assistance systems that can assist the drivers in complex situations. Developing precise driver behavior models near intersections can considerably reduce the number of accidents at road intersections. In this study, the problem of driver behavior modeling near a road intersection is investigated using support vector machines (SVMs) based on the hybrid-state system (HSS) framework. In the HSS framework, the decisions of the driver are represented as a discrete-state system and the vehicle dynamics are represented as a continuous-state system. The proposed modeling technique utilizes the continuous observations from the vehicle and estimates the drivers intention at each time step using a multi-class SVM approach. Statistical methods are used to extract features from continuous observations. This allows for the use of history in estimating the current state. The developed algorithm is trained and tested successfully using naturalistic driving data collected from a sensor-equipped vehicle operated in the streets of Columbus, OH and provided by the Ohio State University. The proposed framework shows a promising accuracy of above 97% in estimating the drivers intention when approaching an intersection.

A probabilistic model of a set of driving decisions

This study proposes a probabilistic decision-making model for driving decisions. The decision-making process that is modeled stochastically is part of the Human Driver Model developed in an earlier study, in which perception, world-model and reflexive behavior are represented as separate modules. Finite-state machine design guidelines for decision-making models are provided to maximize state observability and resolution while maintaining a manageable size for state-machine. Two decision-making models useful for estimation and prediction of driver behavior are presented and one scenario-safety estimation application that uses the proposed decision-making model is given to illustrate the proposed methodology.

An indoor intelligent transportation testbed for urban traffic scenarios

This paper proposes a modular architecture for the development of an indoor testbed for intelligent transportation systems. The main focus is on repeatable, low-cost tests for urban scenarios, especially for higher-level decision making and situation awareness problems. It provides a supplement to outdoor tests and it is also used as a teaching platform. The proposed architecture has been in use at the Ohio State University Control and Intelligent Transportation Research Laboratory, as demonstrated in a number of traffic scenarios.

Trajectory Prediction of a Lane Changing Vehicle Based on Driver Behavior Estimation and Classification

Accurate trajectory prediction of a lane changing vehicle is a key issue for risk assessment and early danger warning in advanced driver assistance systems(ADAS). This paper proposes a trajectory prediction approach for a lane changing vehicle considering high-level driver status. A driving behavior estimation and classification model is developed based on Hidden Markov Models(HMMs). The lane change behavior is estimated by observing the vehicle state emissions in the beginning stage of a lane change procedure, and then classified by the classifier before the vehicle crosses the lane mark. Furthermore, the future trajectory of the lane changing vehicle is predicted in a statistical way combining the driver status estimated by the classifier. The classifier is trained and tested using naturalistic driving data, which shows satisfactory performance in classifying driver status. The trajectory prediction method generates different trajectories based on the classification results, which is important for the design of both autonomous driving controller and early danger warning systems.