Las relaciones de poder entre el Poder Ejecutivo y el poder normativo y fiscalizador regionales – Rep como variable dependiente
5 MARCO DE ATRIBUCIONES Y COMPETENCIAS DE LOS ÓRGANOS REGIONALES
This category includes the use of sensors for positioning, orientation, velocity and vision to detect vehicle motion data. The data is subsequently analysed to find cues to identify irregular driving and then used for some safety related applications such as collision avoidance or lane departure warnings. The following paragraphs discuss the current irregular driving detection related research based on the detection of real-time driving patterns, which includes the generic irregular driving research, irregular driving and collision avoidance and irregular driving and lane control.
Generic irregular driving research
Lecce and Calabrese (2008) proposed an architecture for a driving information system with specific sensors and GPS receivers. The GPS positioning and acceleration data were collected and pattern matching was used to identify and classify driving styles.
able to learn the trajectories and longitudinal/lateral velocities of the vehicle. Then fuzzy neural networks (FNN) are used according to the image processing information to identify whether the vehicle is about to depart the lane or collide with anything ahead. Krajewski et al. (2009), meanwhile, estimated drivers’ fatigue from steering behaviour. The steering data is collected by using orientation sensors and then signal processing is used to capture fatigue impaired patterns. Dai et al. (2010) proposed a system using a mobile phone with an accelerometer and orientation sensor to detect dangerous vehicle manoeuvres typically associated with drunk driving. By computing accelerations based on sensor readings, the user can match the pattern with typical drunk driving patterns extracted from real driving tests to find evidence of drunk driving.
Mohamad et al. (2011) proposed a system for the detection of abnormal driving based on real-time GPS data. The detection algorithm was based on vehicle position and velocity, in particular using maximum speed, maximum acceleration and deceleration as the threshold values. Imkamon et al. (2008) proposed a method for detecting unsafe driving behaviours based on vision and orientation sensors. A fuzzy logic system was used to combine the measured data to classify different levels of hazardous driving. Saruwatari et al. (2012) proposed a method for detecting abnormal driving of vehicles, such as meandering, transverse motion and acceleration or deceleration. In their research, abnormal vehicle motions can be extracted in the sense of group behaviour by using a multi-linear relationship in space-time images. This multi-linear relationship in space-time images could be attained when multiple cameras move with translational motions in different directions at different speeds. Abnormal driving, which is not a translational motion with uniform velocity, can thereby be picked up.
Most of the generic irregular driving research has classified the irregular driving types in a preliminary manner and has not properly quantified the time to first detection, the availability and the correct detection rate of irregular driving. Furthermore, some of the studies used vision sensors where the performance can be affected by the weather conditions. In addition, some studies employing this approach rely significantly on the readings of high grade GPS or other motion sensors, which is not cost effective.
Irregular driving and collision avoidance
Whilst the research reviewed above has focused on the generic detection of irregular driving, the following research works apply the detection of irregular driving to collision-avoidance.
Araki et al. (1996, 1997) developed a collision-avoidance system based on an on- board laser radar and a Charge Coupled Device (CCD) camera, and applied fuzzy logic to evaluate the potential for a collision. The driving status was evaluated by relative distance, relative velocity and the accelerations of both vehicles. Then the risk level of collision was defined based on the comparison with experience data. Risack et al. (2000) developed a lane-keeping assistant video-based system based on vehicle lane position and the time to lane crossing. Their algorithm considered lane geometry and signals from the brakes, steering and indicators. Tsai (2002) also used a laser radar to develop a vehicle safety and warning system and field implementation, while Ueki et al. (2005) developed a vehicular collision-avoidance system by inter-vehicle
crash collision avoidance strategy for cars based on sliding mode control while Tan and Huang (2006) discussed the engineering feasibility of a cooperative collision warning system based on a future trajectory prediction algorithm for vehicles equipped with a differential Global Positioning System (DGPS) unit and other related motion sensors.
Most of the irregular driving and collision avoidance research classifies the risk level in a preliminary way for the collision avoidance application. The research has hardly quantified or even mentioned the performance of different types of irregular driving classification, such as the time to first detection, availability and correct detection rate. Furthermore, most of the research for collision avoidance system uses tightly coupled systems, which compromises compatibility and cost efficiency.
Irregular driving and lane control
This section reviews the research on irregular driving detection related to lane control applications. Kwon et al. (1999) conducted a series of experiments on a vision-based lane departure warning system. The scene in front of the vehicle was captured by a CCD camera. With reference to the steering angle, a perception network approach was applied to build up the decision strategy in order to determine the lane departure of the vehicle. Jung and Kelber (2004, 2005a) developed a Lane Departure Warning System (LDWS) based on a vision system. They designed a linear function to fit the approaching near-vision field, and a quadratic function for the approaching far vision field. They (Jung and Kelber, 2005b) subsequently improved their LDWS, by adding the lateral offset, and its rate of change, from video sequences. Lee and Yi (2005) also
presented a vision based system for lane-departure detection. Departure ratios are introduced to determine instant lane departure and linear regression to minimize false alarms from noise effects. Yoshida et al. (2006) proposed a lane-departure delay strategy for coordinating responses between the vehicle and the driver. The strategy was tested on a driving simulator.
Most of the irregular driving and lane control research has classified irregular driving types in a preliminary and has not properly quantified the performance of irregular driving. Furthermore, most of the studies used vision sensors, whose performance can be affected by the weather and eliminated lane markings. In addition, some of the research employing this approach relies significantly on high grade sensors and/or other complicated auxiliary systems, which are expensive.
Although the real-time driving pattern detection approach has shown significant potential for the detection of irregular driving, research to date is still at an early stage without credible quantification of performance. Furthermore, a technical barrier that needs to be surmounted is the performance of vision sensor based systems that can be affected by adverse weather conditions. In addition, some research employing this approach relies significantly on high grade GPS and other motion sensors, the cost of which may hinder their widespread practical use. Moreover, most of the above- discussed irregular driving detection systems are still at an early stage of development with neither field tests nor a robust algorithm to distinguish different types of irregular driving styles. Their efficiency and reliability need to be further examined, therefore.