Driven Downtown

In a stand-alone mode, no existing navigation technology has the potential to satisfy the requirements for navigation accuracy, continuity, and availability posed by many urban applications. GNSSs generally provide satisfactory performance in open and suburban areas, but have fragmented availability in urban environments due to satellite blockages by buildings and other obstacles. Feature-based navigation shows promise in these areas where enough navigation-related features can be extracted from digital camera or laser scanner images. However, the feature availability can be limited on relatively open streets. A self-contained inertial navigation system (INS) can provide a navigation solution in any environment, but its accuracy drifts over time.

Hence, an integrated multisensor approach that combines complementary features of different navigation techniques can support accurate and reliable navigation capabilities for urban operational scenarios. This article extends earlier development efforts in the areas of INS/laser scanner and GPS/laser scanner integration by combining GPS, laser scanner, and inertial data into a single multisensor integrated solution.

Many existing and prospective applications of navigation systems would benefit from the ability to navigate accurately and reliably in challenging urban environments. Examples include but are not limited to navigation, guidance, and control of autonomous ground vehicles (UGV) and autonomous aerial vehicles (UAV), as well as teams of UGVs and UAVs for urban surveillance and reconnaissance tasks; geographical information system (GIS) urban data collection for mapping applications; monitoring of urban infrastructure for situational awareness; and precise automotive applications such as automated lane-keeping. These applications generally require meter-level to decimeter-level reliable positioning capabilities.

GPS and laser scanner-based navigation approaches ideally complement each other for urban navigation scenarios. The laser scanner-based navigation relies on the availability of structures (lines and surfaces) within the scan range (80 meters, typically). Features such as lines are first extracted from laser scans and then used for position and heading determination. If there is a building wall that blocks GPS signal, this wall provides a feature for the laser-based navigation. Vice versa, for open streets with limited features, the GPS signal is generally unobstructed. Thus, GPS and laser data are combined into an integrated solution architecture. The system architecture developed also exploits INS navigation outputs for improved solution robustness: for example, for robust feature-matching between scan images and for coasting through cases where an insufficient number of combined GPS/laser measurements is available.

A weighted least mean square (LMS) solution procedure estimates changes in user position (or delta position) between successive GPS and laser measurement epochs. In this case, changes in GPS carrier phase and changes in ranges to features extracted from scan images serve as LMS solution observables. Inertial navigation states (position, velocity, and attitude) are applied to predict feature displacements between laser scans in order to improve feature matching, computationally adjust a 2D scan plane for tilting of the laser scanner platform, and coast through cases where not enough combined GPS and laser measurements are available for the delta position estimation. Tight coupling of inertial data with GPS and laser scanner measurements produces a range-domain data fusion carried out by a Kalman filter that utilizes GPS carrier phase and feature ranges to periodically estimate inertial error states.

This article considers a stand-alone GPS receiver option. The algorithms presented are readily modifiable for the differential receiver case.