Primary sensor | Reported performance in research | Sensor cost |
---|---|---|
Camera | 1% level to 0.1% level of travel distance when using visual odometry. For example, 0.46% when navigating a drone indoors on a 75Â m trajectory (with DJI A3 flight integrated camera and Inertial Measurement Unit (IMU)) (Lin et al. 2018); Decimeter-level to centimeter-level when using image matching. For example, lateral and longitudinal location error RMS values of 14.3Â cm and 19.1Â cm in downtown tests, respectively, and 20.5Â cm and 45.4Â cm in parking lot tests (with a Point Grey Flea3 camera and a priori 3D point-cloud database) (Wolcott and Eustice 2014) | Cameras at $ 100 to $ 10 level; Low-cost IMU at $ 100 to $ 10 level |
Camera (infrastructural) | Meter-level to decimeter-level. For example, error mean and standard deviation values of 0.24Â m and 0.19Â m, respectively, when tracking an object in an area of 3Â m by 3Â m in a parking lot (with AVT Prosilica GT 1380 industrial cameras) (Ibisch et al. 2015) | Industrial cameras at $ 1,000 level |
LiDAR | Decimeter-level to centimeter-level. For example, lateral and longitudinal RMS errors of 0.13Â m and 0.16Â m, respectively, when using 3D point cloud matching (with Velodyne HDL-32E LiDAR) (Wolcott and Eustice 2017) | 2D LiDAR in $1,000 to $100 level; 3D LiDAR in $ 10,000 to $ 1,000 level |
LiDAR (infrastructural) | Meter-level to decimeter-level. For example, mean absolute lateral and longitudinal errors of 6.3Â cm and 8.5Â cm, respectively (with Sick LMS500 LiDARs) (Ibisch et al. 2013) | 2D LiDAR in $1,000 to $100 level; 3D LiDAR in $ 10,000 to $ 1,000 level |
HD map | Decimeter-level to centimeter-level. For example, lateral and longitudinal mean location errors of 14Â cm and 26Â cm, respectively (use HD map to enhance solution from two BlackFly PGE-50S5M cameras, a Ublox C94-M8P Global Navigation Satellite System (GNSS) receiver, and a vehicle odometer) (Poggenhans et al. 2018) | $ 1,000 level per vehicle (without 3D LiDAR); $ 10,000 level per vehicle (with 3D LiDAR); significant extra cost for HD-map maintenance |
Inertial sensors (vehicle) | 1% level of travel distance. For example, 1.10% of travel distance in driving of 800Â m (with RT30032 IMU) (Brossard et al. 2020) | Low-cost IMU at $ 100 to $ 10 level |
Inertial sensors (pedestrian) | 1% level of travel distance. For example, 0.5% of travel distance in walking of 800Â m (with 4 InvenSense IMUs on both feet) (Niu et al. 2019) | Low-cost IMU at $ 100 to $ 10 level |
Visible light | Decimeter-level to centimeter-level. For example, an RMS error of 20Â cm (with 5 OPT101 photodiodes in an area of 5Â m by 5Â m) (Zhuang et al. 2019) | Infrastructure cost $ 100 level per 100 m2-level area |
Ultra-Wide Band (UWB)/ultrasonic/pseudolite | Decimeter-level to centimeter-level. For example, an RMS error within 10Â cm and 95% within 20Â m (with 8 UWB anchors in an area of 7Â m by 7Â m) (Tiemann et al. 2015) | Infrastructure cost $ 100 level per 100 m2-level area |
Wireless Fidelity (WiFi)/Bluetooth Low Energy (BLE) | Mete-level when using Received Signal Strength (RSS). For example, 90% within 3.88Â m (with 8 BLE beacons in a 40Â m by 60Â m area (Zhuang et al. 2016); Decimeter-level to centimeter-level when using Time-of-Arrival (ToA), Angle-of-Arrival (AoA), Channel-State-Information (CSI), or Round-Trip-Time (RTT). For example, a median of 4.17Â cm (with 1 WiFi access point in a 6Â m by 8Â m area) (Vasisht et al. 2016) | Infrastructure cost of $ 10 level per 100 m2-level area; extra cost for infrastructure modification when using ToA, AoA, CSI, or RTT |
The fifth generation of mobile network communication technology (5G) | 100-m-level to centimeter-level, depending on the density of base stations and measurement used. For example, centimeter-level in (Witrisal et al. 2016). Most current 5G PLAN research is based on theoretical analysis or simulation data | $1000 to $ 100 level per base station. Coverage range from kilometer-level to within 100Â m |