Alberto Yukinobu Hata

CaRINA Intelligent Robotic Car: Architectural design and applications

Abstract This paper presents the development of two outdoor intelligent vehicles platforms named CaRINA I and CaRINA II, their system architecture, simulation tools, and control modules. It also describes the development of the intelligent control system modules allowing the mobile robots and vehicles to navigate autonomously in controlled urban environments. Research work has been carried out on tele-operation, driver assistance systems, and autonomous navigation using the vehicles as platforms to experiments and validation. Our robotic platforms include mechanical adaptations and the development of an embedded software architecture. This paper addresses the design, sensing, decision making, and acting infrastructure and several experimental tests that have been carried out to evaluate both platforms and proposed algorithms. The main contributions of this work is the proposed architecture, that is modular and flexible, allowing it to be instantiated into different robotic platforms and applications. The communication and security aspects are also investigated.

Robust curb detection and vehicle localization in urban environments

Curb detection is an important capability for autonomous ground vehicles in urban environments. It is particularly useful for path planning and safe navigation. Another important task that can benefit from curb detection is localization, which is also a major requirement for self-driving cars. There are several approaches for identifying curbs using stereo cameras and 2D LIDARs in the literature. Stereo cameras depend on image pair matching methods to obtain depth information. Although 2D LIDARs being able to directly return this information, only few curb points can be detected using this sensor. In this work we propose the use of a 3D LIDAR which provides a dense point cloud and thus make possible to detect a larger extent of the curb. Our approach introduces the use of robust regression method named least trimmed squares (LTS) to deal with occluding scenes in contrast of temporal filters and spline fitting methods. We also used the curb detector as an input of a Monte Carlo localization algorithm, which is capable to estimate the pose of the vehicle without an accurate GPS sensor. We conducted experiments in urban environments to validate both the curb detector and the localization algorithm. Both method delivered successful results in different traffic situations and an average lateral localization error of 0.52655 m in a 800 m track.

Road marking detection using LIDAR reflective intensity data and its application to vehicle localization

A correct perception of road signalizations is required for autonomous cars to follow the traffic codes. Road marking is a signalization present on road surfaces and commonly used to inform the correct lane cars must keep. Cameras have been widely used for road marking detection, however they are sensible to environment illumination. Some LIDAR sensors return infrared reflective intensity information which is insensible to illumination condition. Existing road marking detectors that analyzes reflective intensity data focus only on lane markings and ignores other types of signalization. We propose a road marking detector based on Otsu thresholding method that make possible segment LIDAR point clouds into asphalt and road marking. The results show the possibility of detecting any road marking (crosswalks, continuous lines, dashed lines). The road marking detector has also been integrated with Monte Carlo localization method so that its performance could be validated. According to the results, adding road markings onto curb maps lead to a lateral localization error of 0.3119 m.

Feature Detection for Vehicle Localization in Urban Environments Using a Multilayer LIDAR

Localization is an important component of autonomous vehicles, as it enables the accomplishment of tasks, such as path planning and navigation. Although vehicle position can be obtained by GNSS devices, they are susceptible to errors and satellite signal unavailability in urban scenarios. Several map-aided localization solution methods have been proposed in the literature, but mostly for indoor environments. Maps used for localization store relevant environmental features that are extracted by a detection method. However, many feature detection methods do not consider the presence of dynamic obstacles or occlusions in the environment, which can impair the localization performance. In order to detect curbs even in occluding scenes, we developed a method based on ring compression analysis and least trimmed squares. For road marking detection, we developed a modified version of the Otsu thresholding method to segment road painting from road surfaces. Finally, the feature detection methods were integrated with a Monte Carlo localization method to estimate the vehicle position. Experimental tests in urban streets have been used to validate the proposed approach with favorable results.

Artificial Neural Nets Object Recognition for 3D Point Clouds

This paper presents a approach that uses 3D point clouds from laser sensor to perform the classification of typical obstacles in urban environments. The presented method consists of Velodyne Lidar point clouds segmentation, feature extraction and use of a MLP Neural Nets to classify vehicles, people, tree trunks, light poles and buildings. Experimental results demonstrated that is possible to recognize different classes of 3D structures with a very good precision. At the end, the performances of two neural networks are compared.

Intelligent control and evolutionary strategies applied to multirobotic systems

This paper describes the modeling, implementation, and evaluation of RoBombeiros multirobotic system. The robotic task in this paper is performed over a natural disaster, simulated as a forest fire. The simulator supports several features to allow realistic simulation, like irregular terrains, natural processes (e.g. fire, wind) and physical constraint in the creation and application of mobile robots. The proposed system relies on two steps: (i) group formation planning and (ii) intelligent techniques to perform robots navigation for fire fighting. For planning, we used genetic algorithms to evolve positioning strategies for firefighting robots performance. For robots operation, physically simulated fire-fighting robots were built, and the sensory information of each robot (e.g. GPS, compass, sonar) was used in the input of an artificial neural network (ANN). The ANN controls the vehicle (robot) actuators and allows navigation with obstacle avoidance. Simulation results show that the ANN satisfactorily controls the mobile robots; the genetic algorithm adequately configures the fire fighting strategy and the proposed multi-robotic system can have an essential hole in the planning and execution of fire fighting in real forests.

Crossroad Detection Using Artificial Neural Networks

An autonomous ground vehicle has to be able to execute several tasks such as: environment perception, obstacle detection, and safe navigation. The road shape provides essential information to localization and navigation. It can also be used to identify reference points in the scenario. Crossroads are usual road shapes in urban environments. The detection of these structures is the main focus of this paper. Whereas cameras are sensible to illumination changes, we developed methods that handle LIDAR (Light Detection And Ranging) sensor data to accomplish this task. In the literature, neural networks have not been widely adopted to crossroad detection. One advantage of neural networks is its capability to deal with noisy data, so the detection can be performed even in the presence of other obstacles as cars and pedestrians. Our approach takes advantage of a road detector system that produces curb data and road surface data. Thus we propose a crossroad detector that is performed by an artificial neural network and LIDAR data. We propose two methods (curb detection and road surface detection) for this task. Classification results obtained by different network topologies have been evaluated and the performance compared with ROC graphs. Experimental tests have been carried out to validate the approaches proposed, obtaining good results when compared to other methods in the literature.

Outdoor Mapping Using Mobile Robots and Laser Range Finders

This paper describes a three-dimensional mapping method to allow the operation of mobile robots in outdoor environments using 2D laser range finders (LRF). Experimental tests demonstrate the accuracy of the presented techniques, allowing for precise models of the environments traversed by the robot. This work also presents preliminary results of terrain classification according to its navigability.

3D Point Clouds Segmentation for Autonomous Ground Vehicle

Point clouds segmentation is an essential step to improve the performance of obstacle detection and classification in areas of autonomous ground vehicles and mobile robotics. This paper presents a study and comparison of the performance of segmentation methods using point clouds coming from a 3D laser sensor, more specifically obtained from a Velodyne HDL32.

Terrain mapping and classification using Support Vector Machines

This paper describes a three-dimensional terrain mapping and classification technique to allow the operation of mobile robots in outdoor environments using laser range finders. We use Support Vector Machines to classify portions of mapped terrain into navigable, partially navigable and non-navigable. In order to detect safe places to robot traverse, our approach can be used to assist the robot navigation in unstructured lands. Experimental results obtained using real environments and robot show the efficiency of the presented methods.