RESUMO
In the near future, the communication between autonomous cars will produce a network of sensors that will allow us to know the state of the roads in real time. Lidar technology, upon which most autonomous cars are based, allows the acquisition of 3D geometric information of the environment. The objective of this work is to use point clouds acquired by Mobile Laser Scanning (MLS) to segment the main elements of road environment (road surface, ditches, guardrails, fences, embankments, and borders) through the use of PointNet. Previously, the point cloud was automatically divided into sections in order for semantic segmentation to be scalable to different case studies, regardless of their shape or length. An overall accuracy of 92.5% has been obtained, but with large variations between classes. Elements with a greater number of points have been segmented more effectively than the other elements. In comparison with other point-by-point extraction and ANN-based classification techniques, the same success rates have been obtained for road surfaces and fences, and better results have been obtained for guardrails. Semantic segmentation with PointNet is suitable when segmenting the scene as a whole, however, if certain classes have more interest, there are other alternatives that do not need a high training cost.
RESUMO
Objective: In this work, a roundabout and a turbo roundabout model are compared and previous modeling with continuous Petri nets and safety are analyzed through indicators of complexity. Petri nets are a graphic and mathematical representation that allow a faithful modeling of urban systems. Method: The methodology has been designed for the transformation of a real system to small subgraphs that represent the maneuvers in roundabouts, approximated as roads and lanes of incorporation. Places within the roundabout have been located and defined as continuous places from their influence and visibility toward adjacent conditions. The transitions have been modeled by time and inhibitory arcs, which represent priorities and areas where drivers must pay attention. The created networks represent a faithful model of vehicle flow trajectories in the roundabouts. Results: The methodology is applied to the same real road intersection. The case study is a recent transformation from roundabout to turbo roundabout. The roundabout network complexity is corroborated by a greater number of entries and exits that lead to each roundabout place (reflected in the maneuvers that can be performed) and a greater number of inhibiting arcs. In most of the turbo roundabout places, the driver's only option is reduced to occupying next place. The possibility of choosing between several places supposes a greater trajectory intersection and an increased time for decision making. The only situation where the complexity is the same between both systems is when a vehicle accesses the inner lane of the roundabout from the left lane on a single-lane road. The main maneuvers causing accidents have been modeled and their solution in a turbo roundabout is presented. Conclusions: The reduced complexity of the turbo roundabout is due to the strict limitations in lane changes, turning turbo roundabouts into a safer model: A lower number of possible movements that can be performed by drivers and a smaller number of trajectories with collision risk. Petri nets have proven to be perfectly applicable to the representation of traffic circular systems (such as roundabouts and turbo roundabouts) and to measure the complexity and security of the system.