Raúl Quintero

Free space and speed humps detection using lidar and vision for urban autonomous navigation

In this paper, a real-time free space detection system is presented using a medium-cost lidar sensor and a low cost camera. The extrinsic relationship between both sensors is obtained after an off-line calibration process. The lidar provides measurements corresponding to 4 horizontal layers with a vertical resolution of 3.2 degrees. These measurements are integrated in time according to the relative motion of the vehicle between consecutive laser scans. A special case is considered here for Spanish speed humps, since these are usually detected as an obstacle. In Spain, speed humps are directly related with raised zebra-crossings so they should have painted white stripes on them. Accordingly the conditions required to detect a speed hump are: detect a slope shape on the road and detect a zebra crossing at the same time. The first condition is evaluated using lidar sensor and the second one using the camera.

Pedestrian path prediction using body language traits

Driver Assistance Systems have achieved a high level of maturity in the latest years. As an example of that, sophisticated pedestrian protection systems are already available in a number of commercial vehicles from several OEMs. However, accurate pedestrian path prediction is needed in order to go a step further in terms of safety and reliability, since it can make the difference between effective and non-effective intervention. In this paper, we consider the three-dimensional pedestrian body language in order to perform path prediction in a probabilistic framework. For this purpose, the different body parts and joints are detected using stereo vision. We propose the use of GPDM (Gaussian Process Dynamical Models) for reducing the high dimensionality of the input feature vector (composed by joints and displacement vectors) in the 3D pose space and for learning the pedestrian dynamics in a latent space. Experimental results show that accurate path prediction can be achieved at a time horizon of ≈ 0.8 s.

Pedestrian Path Prediction Based on Body Language and Action Classification

Safety-related driver assistance systems are becoming mainstream and nowadays many automobile manufacturers include them as standard equipment. For example, pedestrian protection systems are already available in a number of commercial vehicles. However, there is still work to do in the improvement of the accuracy of these systems since the difference between an effective and a non-effective intervention can depend on a few centimeters or on a fraction of a second. In this paper, we use the 3D pedestrian body language in order to perform accurate pedestrian path prediction by means of action classification. To carry out the prediction, we propose the use of GPDM (Gaussian Process Dynamical Models) that reduces the high dimensionality of the input vector in the 3D pose space and learns the pedestrian dynamics in a latent space. Instead of combining a reduced number of subjects in a single model that will have to deal with the stylistic variations, we propose a much more scalable approach where all the subjects are separately trained in individual models. These models will be then hierarchically separated according to their action (walking, starting, standing, stopping) and direction of the motion. Finally, for a test sequence, the appropiate model will be selected by means of an action classification system based on the similarity of the 3D poses transitions and the joints velocities. The estimated action will constrain the models to use for the prediction, taking into account only the ones trained for that action. Experimental results show that the system has the potential to provide accurate path predictions with mean errors of 7 cm, for walking trajectories, 20 cm, for stopping trajectories and 14 cm for starting trajectories, at a time horizon of 1 s.

Pedestrian Intention and Pose Prediction through Dynamical Models and Behaviour Classification

Pedestrian protection systems are being included by many automobile manufacturers in their commercial vehicles. However, improving the accuracy of these systems is imperative since the difference between an effective and a non-effective intervention can depend only on a few centimeters or on a fraction of a second. In this paper, we describe a method to carry out the prediction of pedestrian locations and pose and to classify intentions up to 1 s ahead in time applying Balanced Gaussian Process Dynamical Models (B-GPDM) and naïve-Bayes classifiers. These classifiers are combined in order to increase the action classification precision. The system provides accurate path predictions with mean errors of 24.4 cm, for walking trajectories, 26.67 cm, for stopping trajectories and 37.36 cm for starting trajectories, at a time horizon of 1 second.

Stereo-based pedestrian detection in crosswalks for pedestrian behavioural modelling assessment

In this paper, a stereo- and infrastructure-based pedestrian detection system is presented to deal with infrastructure-based pedestrian safety measurements as well as to assess pedestrian behaviour modelling methods. Pedestrian detection is performed by region growing over temporal 3D density maps, which are obtained by means of stereo reconstruction and background modelling. 3D tracking allows to correlate the pedestrian position with the different pedestrian crossing regions (waiting and crossing areas). As an example of an infrastructure safety system, a blinking luminous traffic sign is switched on to warn the drivers about the presence of pedestrians in the waiting and the crossing regions. The detection system provides accurate results even for nighttime conditions: an overall detection rate of 97.43% with one false alarm per each 10 minutes. In addition, the proposed approach is validated for being used in pedestrian behaviour modelling, applying logistic regression to model the probability of a pedestrian to cross or wait. Some of the predictor variables are automatically obtained by using the pedestrian detection system. Other variables are still needed to be labelled using manual supervision. A sequential feature selection method showed that time-to-collision and pedestrian waiting time (both variables automatically collected) are the most significant parameters when predicting the pedestrian intent. An overall predictive accuracy of 93.10% is obtained, which clearly validates the proposed methodology.

Assistive Pedestrian Crossings by Means of Stereo Localization and RFID Anonymous Disability Identification

Assistive technology usually refers to systems used to increase, maintain, or improve functional capabilities of individuals with disabilities. This idea is here extended to transportation infrastructures, using pedestrian crossings as a specific case study. We define an Assistive Pedestrian Crossing as a pedestrian crossing able to interact with users with disabilities and provide an adaptive response to increase, maintain or improve their functional capabilities while crossing. Thus, the infrastructure should be able to locate the pedestrians with special needs as well as to identify their specific disability. In this paper, user location is obtained by means of a stereo-based pedestrian detection system. Disability identification is proposed by means of a RFID-based anonymous procedure from which pedestrians are only required to wear a portable and passive RFID tag. Global nearest neighbor is applied to solve data association between stereo targets and RFID measurements. The proposed assistive technology is validated in a real crosswalk, including different complex scenarios with multiple RFID tags.

Monocular vision-based target detection on dynamic transport infrastructures

This paper describes a target detection system on transport infrastructures, based on monocular vision. The goal is to detect and track vehicles and pedestrians, dealing with objects variability, different illumination conditions, shadows, occlusions and rotations. A background subtraction method, based on GMM and shadow detection algorithms are proposed to do the segmentation of the image. Finally a feature extraction, optical flow analysis and clustering methods are used for the tracking step. The algorithm requires no object model and prior knowledge and it is robust to illumination changes and shadows.

Recognizing individuals in groups in outdoor environments combining stereo vision, RFID and BLE

Vision-based people localization systems in outdoor environments can be enhanced by means of radio frequency identification technologies. This combination has the potential to enable a wide range of new applications. When individuals wear a radio frequency tag, they may be both identified and localized. In this way, the technology may interact with individuals in a personalized way. In this paper, two radio frequency identification technologies, UHF Radio Frequency IDentification (RFID) and Bluetooth Low Energy (BLE), are combined with a stereo-based people detection system to recognize individuals in groups in complex outdoor scenarios in medium sized areas up to 20 m. The proposed approach is validated in crosswalks with pedestrians wearing portable RFID passive tags and active BLE beacons.

Extended Floating Car Data system - experimental study

This paper presents the results of a set of extensive experiments carried out in daytime and nighttime conditions in real traffic using an enhanced or extended Floating Car Data system (xFCD) that includes a stereo vision sensor for detecting the local traffic ahead. The detection component implies the use of previously monocular approaches developed by our group in combination with new stereo vision algorithms that add robustness to the detection and increase the accuracy of the measurements corresponding to relative distance and speed. Besides the stereo pair of cameras, the vehicle is equipped with a low-cost GPS and an electronic device for CAN Bus interfacing. The xFCD system has been tested in a 198-minutes sequence recorded in real traffic scenarios with different weather and illumination conditions, which represents the main contribution of this paper. The results are promising and demonstrate that the system is ready for being used as a source of traffic state information.

Surface classification for road distress detection system enhancement

This paper presents a vision-based road surface classification in the context of infrastructure inspection and maintenance, proposed as stage for improving the performance of a distress detection system. High resolution road images are processed to distinguish among surfaces arranged according to the different materials used to build roads and their grade of granulation and striation. A multi-class Support Vector Machine (SVM) classification system using mainly Local Binary Pattern (LBP), Gray-Level Co-occurrence Matrix (GLCM) and Maximally Stable Extremal Regions (MSER) derived features is described. The different texture analysis methods are compared based on accuracy and computational load. Experiments with real application images show a significant improvement on the the distress detection system performance by combining several feature extraction methods.