Intelligent Systems LabUniversidad Carlos III de Madrid
HomePeopleResearchPublicationsMediaContact

Technologies

Stereo Based Obstacle Detection

    Intelligent Vehicles demand a trustable 3D environment perception. In order to provide full environment perception LSI developed a stereo camera based obstacle detection, able to provide free space detection and to identify objects in the road plane. The algorithm is based on the use of u-v disparity algorithms and it is able to identify objects with vertical projection, furthermore it is able to identify y those that represent potential risk for the vehicle.

Platforms: IVVI 2.0 & iCab

Infrared Pedestrian Detection

    The infrared detection system, developed by LSI, is based on the hope-features that are specifically adapted for infrared technologies. The use of shape-based features allows the system to work under strong heat condition, i.e. during day-light.

Platforms: IVVI 2.0 & iCab

Camera-based Detection

    Different detection approaches were developed in LSI, using the most recent techniques of computer vision, in order to provide trustable visual detection. The different techniques allowed to develop high quality visual applications able to be used in the most challenging scenarios. The algorithms developed were adapted to both stereo vision systems and monocular systems.

  • Pedestrian Detection
  • Vehicle Detection
  • Motorcycle Detection
  • Cyclist Detection (under development)

Platforms: IVVI 2.0 & iCab

Detection based on Data Fusion

    Detection based on laser scanner and stereo cameras are combined to enhance classical detection. The accuracy of the location and credibility of obstacle detection from laser scanner is combined with the camera. The data from the camera allows to provide extra information, which allows obstacle classification.

  • Pedestrian Detection
  • Vehicle Detection
  • Motorcycle Detection
  • Cyclist Detection (under development)

Platforms: IVVI 2.0 & iCab

Traffic Signs Detection & Identification

    Traffic sign detection is one of the first technologies developed by LSI, the system is able to work under the most extreme condition, including changing visibility condition, vandalism and occlusions.

Platforms: IVVI 2.0 & iCab

Traffic Lanes Detection & Identification

    A stereo-vision algorithm was developed by LSI allows the extraction of number, position and type (e.g. acceleration) of road lanes starting from a three-dimensional environment model. Ego-lane, the one at which the vehicle is traveling, is also detected; as well as the deviation of the vehicle from the center of the lane. Therefore, the surrounding scene can be fully understood from the car's point of view.

    Moreover, line classification is performed by means of machine learning technologies and a temporal filtering is also implemented to ensure robust estimation.

Platforms: IVVI 2.0 & iCab

Visual Odometry

    Visual odometry is a state of the art technology, based on the use of features detected in the image, 3D information from the stereo camera and an advanced tracking algorithm. The algorithms designed are able to track the movement of the vehicle, providing vehicle localization, even in the event of GPS signal loss. This technology is implemented for ground vehicles, based on stereo camera and in aerial vehicles based on monocular camera and embedded sensors.

Platforms: IVVI 2.0iCab & Drones

Laser Scanner & Camera Data Fusion

    Data fusion algorithms combining information from Lidar scanners and regular cameras require an accurate calibration in order to provide reliable information. Extrinsic parameters relating both sensors, namely translation and rotation, can be automatically extracted by means of an algorithm with narrow requirements. Such as only the road plane and an obstacle, of any shape, are assumed to be simultaneously visible by both sensors, enabling a calibration process that is performed without any human intervention.

Platforms: IVVI 2.0 & iCab

Autonomous Navigation

    An autonomous vehicle, either ground or aerial, is a vehicle that performs behaviors or tasks with a high degree of autonomy, at which it can gain information about the environment, work for an extended period of time without human intervention, move itself throughout its operating environment without human assistance and avoid situations that are harmful to people, property, or itself.

Platforms: Drones & iCab

Driver Monitoring

   Driver Monitoring technologies are based in two different approaches which are currently being merged in a high lever monitoring application. First, is based on the used on perception technologies, such as monocular cameras, time of flight cameras and stereo systems to provide driver monitoring. By means of advanced 2D and 3D algorithms head pose, eye gaze and eye blinking are detected, providing distraction and drowsiness detection. The second approach is based on the use of in-vehicle signals (acquired through CAN-Bus) and other embedded sensors, such as GPS and inertial measurements, to identify aggressive driver maneuvers. By the fusion of both technologies, it is possible to provide advance driver monitoring system, based on advance fusion techniques.

Platforms: IVVI 2.0

Pedestrian & Vehicle Tracking

    Tracking algorithms were developed to enhance the detection by means of strong tracking techniques and advanced data association algorithms.

    The tracking algorithms include Kalman Filter for visual odometry and object tracking, Extended Kalman Filter for data fusion and GPS enhancement, Unscented Kalman Filter for visual odometry and object tracking, Particle Filter for object tracking and PHD Filter for object tracking.

    The developed association techniques include Multiple Hypothesis Tracking, Nearest Neighbors and Joint Probabilistic Data Fusion.

Platforms: IVVI 2.0 & iCab

Advanced GPS Localization

    By means of data fusion techniques, the accuracy of GPS is enhanced by fusing inertial measurement unit information and GPS signals. Furthermore, visual odometry and other positioning techniques are fused together to provide accurate localization in absence of high quality signalling.

Platforms: IVVI 2.0 & iCab

CAN Bus Monitoring

    CAN Bus analysis software and hardware was developed by LSI. The hardware is able to communicate in real-time with the vehicle. The included micro-controller and LCD display allows an online configuration in different modes: Online data retrieval, data storage and online communication analysis.

Platforms: IVVI 2.0

UAV Autonomous Landing

    This technology provides extension to the autonomous navigation algorithms developed for aerial vehicles. In particular, it deals with the landing process. By the use of autonomous control techniques and computer vision, the drone navigates the surroundings until it finds the designed pattern, then apply maneuvers to land over the pattern.

Platforms: Drones
 

People Detection from UAV

    This technology deals with a people detection and counting algorithm using a camera mounted on the drone. The UAVs solve the problem of large number of cameras in large areas. The implemented detection algorithm uses Histogram of Oriented Gradients (HOG) approach and Support Vector Machine (SVM) classifier. The algorithm uses a combination of body and head detection approaches to enhance the counting process in high altitudes and crowded areas.

Platforms: Drones
 

Obstacle Detection from UAV

    This technology implements an obstacle detection algorithm that mimics human's behavior of detecting the state of approaching obstacles using single camera mounted on the drone. During flight, the method detects and estimates the changes of the size area of the obstacles. First, the method detects the feature points of the obstacles, then extracts the obstacles that has probability of getting close. In addition, by comparing the area ratio of the obstacle and the position of the drone, the method can decide if it is obstacle or not. Finally, by estimating the obstacle 2D position in the image and combining with the tracked waypoints, the UAV can take the action of avoidance.

Platforms: Drones
 

Mobile Phone Apps

    Several of the available technologies are already available for android mobile phones. Including applications of traffic sign detection, driver monitoring, vehicle overtaking detection, vehicle tracking and more.