Hybrid Map-Based Navigation Method for Unmanned Ground Vehicle in Urban Scenario
To reduce the data size of metric map and map matching computational cost in unmanned ground vehicle self-driving navigation in urban scenarios, a metric-topological hybrid map navigation system is proposed in this paper. According to the different positioning accuracy requirements, urban areas are divided into strong constraint (SC) areas, such as roads with lanes, and loose constraint (LC) areas, such as intersections and open areas. As direction of the self-driving vehicle is provided by
... fic lanes and global waypoints in the road network, a simple topological map is fit for the navigation in the SC areas. While in the LC areas, the navigation of the self-driving vehicle mainly relies on the positioning information. Simultaneous localization and mapping technology is used to provide a detailed metric map in the LC areas, and a window constraint Markov localization algorithm is introduced to achieve accurate position using laser scanner. Furthermore, the real-time performance of the Markov algorithm is enhanced by using a constraint window to restrict the size of the state space. By registering the metric maps into the road network, a hybrid map of the urban scenario can be constructed. Real unmanned vehicle mapping and navigation tests demonstrated the capabilities of the proposed method. OPEN ACCESS Remote Sens. 2013, 5 3663 Keywords: hybrid map; unmanned ground vehicle; topological map; metric map; simultaneous localization and mapping; Markov localization; laser scanner Introduction Autonomous-driving technologies have attracted considerable academic and industrial interests in recent years. Navigating an unmanned ground vehicle (UGV) driving through urban areas is a difficult task. The UGV must know exactly where it is in dynamic traffic scenarios. The accuracy of the location information directly impacts the safety and control stability of the UGV. The localization methods of autonomous driving can be divided into satellite signal-based solutions and map matching solutions. In satellite signal-based solutions, GPS is the most common positioning way. However, because of the multipath effect and the occlusion of satellite signals caused by buildings, trees or clouds, the ordinary GPS equipment with normal accuracy cannot meet the requirement of the positioning system. Therefore, the differential GPS (DGPS) or GPS-inertial navigation system (INS) positioning system comes into use. In the 2007 Defense Advanced Research Projects Agency (DARPA) Urban Challenge, most of the unmanned vehicles were equipped with costly DGPS-based navigation systems [1,2]. The expensive equipment will increase the cost of the autonomous driving vehicle. When in a big city, DGPS positioning signals cannot cover everywhere. Moreover, when there are high buildings, big trees, bridges or tunnels, it is impossible for the positioning system to obtain even ordinary GPS satellite signals. For the map matching-based method, an unmanned vehicle gets its position by comparing the environment map with real-time perception data from mounted sensors, such as laser scanners or cameras      . If the map is large and accurate enough, map matching-based approaches show better stability than the GPS-based method in the urban scenario. Such an environment map can be obtained by various approaches    . However, the tremendous size and the huge cost of a detailed map of a whole city is a big problem. Storage and computational complexity, due to a massive amount of map data, will affect the efficiency of positioning system. In early 2013, the Google self-driving car was reported to be commercially available to customers in five to seven years  . However, how to obtain the accurate position of the self-driving car is still a challenge. In this kind of map matching approach, the determination of the autonomous vehicle relies on very detailed maps of the roads and terrain. Before sending the self-driving car on a road test, a human driver must drive along the route to gather data about the environment. Then, the autonomous vehicle can compare the data acquired from the perception sensors to the previously recorded map data  . Therefore, a very detailed, accurate and high-cost digital map along the driving routes is essential. The high cost refers to not only the map building, but the calculation load of the matching process during autonomous driving. From the experiences of participation in the Intelligent Vehicle Future Challenge competitions of China since 2009 , we found that a UGV could take advantage of a variety of positioning methods, rather than a particular one in the entire self-driving journey in urban areas. In most cases, the UGV drives on the structured roads with lanes and does not need precise position data. Lanes, road curbs and global path points ahead provide the direction for the UGV that drives itself like water flows in a pipe.