OSM-based navigation and decision-making
J. Godoy, A. Artuñedo, J. VIllagra
Description of the technology
AUTOPIA works on decision-making for cooperative and automated vehicles. The research group aims at pushing the navigation capabilities of these vehicles when only low-cost and open-source maps are available. To that end, self-generated drivable corridors, based on Open Street Maps, are dynamically obtained. To cope with their intrinsic uncertainty and low-fidelity, a camera-based lane detection system updates and enhances the navigable space. From that point, an efficient and human-like traffic-free local planner determines, under a probabilistic framework, a safe motion primitive. Multi-sensor perception is then used to identify the driving scenario, and eventually re-plan the local path and speed profile, leading in some cases to re-adapt the maneuver or route to be followed.
The whole software architecture is based on the open-source Lightweight Communication and Marshalling (LCM).
Experimental set up & Demonstration
The experimental platform is a fully automated Citroën DS3, equipped with RTK-DGPS, IMU, stereovision and multi-layer Lidars. The demonstration will showcase the navigation architecture of AUTOPIA, ranging dynamic global planning to local motion planning, including obstacle avoiding. An on-board Android human-machine interface relying on OsmAnd will be able to (i) compute the optimal routing to be followed under different scenarios, (ii) show the most relevant information perceived by the embedded sensors. From this high-level mission, the vehicle will dynamically generate a human and efficient path and speed profile. In the event a static or dynamic obstacle is detected on the road by a multi-layer LiDAR, the planner will evaluate whether an avoiding maneuver is feasible in the pre-computed route, or an automatic re-routing request is performed.
 A. Artuñedo, J. Godoy and J. Villagra, A Primitive Comparison for Traffic-Free Path Planning, in IEEE Access, vol. 6, pp. 28801-28817, 2018. [PDF]
 J. Godoy, A. Artuñedo, and J. Villagra, Self-generated OSM-based driving corridors, submitted to IEEE Access, 2018
 A. Artuñedo, J. Godoy and J. Villagra, Smooth path planning for urban autonomous driving using OpenStreetMaps, 2017 IEEE Intelligent Vehicles Symposium (IV), Los Angeles, CA, 2017, pp. 837-842. [PDF]
 J. Godoy, J. Pérez, E. Onieva, J. Villagrá, V. Milanés and R. Haber, A driverless vehicle demonstration on motorways and in urban environments, Transport, vol. 30, no. 3, 253-263, 2016 [PDF]