LaneQuest: An accurate and energy-efficient lane detection system

Abstract

Abstract-Current outdoor localization techniques fail to provide the required accuracy for estimating the car's lane. In this paper, we present LaneQuest: a system that leverages the ubiquitous and low-energy inertial sensors available in commodity smart-phones to provide an accurate estimate of the car's current lane. LaneQuest leverages hints from the phone sensors about the surrounding environment to detect the car's lane. For example, a car making a right turn most probably will be in the right-most lane, a car passing by a pothole will be in a specific lane, and the car's angular velocity when driving through a curve reflects its lane. Our investigation shows that there are amble opportunities in the environment, i.e. lane "anchors", that provide cues about the car's lane. To handle the ambiguous location, sensors noise, and fuzzy lane anchors; LaneQuest employs a novel probabilistic lane estimation algorithm. Furthermore, it uses an unsupervised crowd-sourcing approach to learn the position and lane-span distribution of the different lane-level anchors.Our evaluation results from implementation on different android devices and 260Km driving traces by 13 drivers in different cities shows that LaneQuest can detect the different lane-level anchors with an average precision and recall of more than 90%. This leads to an accurate detection of the exact car's lane position 80% of the time, increasing to 89% of the time to within one lane. This comes with a low-energy footprint, allowing LaneQuest to be implemented on the energy-constrained mobile devices.

Parkmaster

“…In this case, car detection must be enabled only when the user is driving on the rightmost lane. While at the moment ParkMaster assumes the user is always doing so, we believe vision-based [40] or sensors-based [6,7,10,15,43] solutions for lane detection can be easily integrated with ParkMaster to trigger the detection only when the user is driving on the right lane.…”

semMatch

“…To estimate σ h , we use the ground-truth data. Note that, unlike the input positioning data which has highly dynamic and variable accuracy, the sensors heading accuracy has a more consistent accuracy outdoors [4]. Hence, we use a fixed σ h calculated based on the Median Absolute Deviation (MAD) of the direction changes in our ground truth data, which is a robust estimator of σ h [4,28]:…”

Carloc

“…CARLOC's use of vehicle sensors and crowd-sourced landmarks, together with advanced map matching, gives it an order of magnitude higher accuracy than the prior work. LaneQuest [2] uses probabilistic methods to estimate which lane a car is driving on, a qualitatively different problem than ours. LaneQuest, however, uses crowd-sourced anchors, but, unlike CARLOC, cannot leverage vehicle sensors to detect these.…”