Search

For an autonomous vehicle, reliable situation understanding is a key component of safe navigation. An incorrect prediction of an upcoming situation means that erroneous information may be supplied to the decision-making process, leading to hazardous outcomes. It is therefore of great importance to estimate the driving areas that are reachable by other interacting road users, without introducing misleading information. This paper presents a means of handling the integrity of prediction information, given the imperfection of object prediction, via a Lane Grid Map, that is to say a spatial representation of the situation at a tactical level, based on the topological layer of a high-definition map. We demonstrate experimentally, using real data, how the spatial sampling step of the grid representation can be used to manage the integrity of prediction information. Moreover, addressing interactions during the prediction makes it possible to handle some particular situations safely. We show how some interactions can be utilized via the concept of neutralization. To quantify the integrity of the prediction, we propose the use of two metrics, namely False Negative Rate and Neutralized Time Interval. Experiments were carried out with three vehicles to evaluate the integrity of the prediction using these metrics.

Navigation maps provide critical information for advanced driving assistance systems and autonomous vehicles. When these maps are refined to lane-level, ambiguities may occur during the map-matching process, particularly when positioning estimates are inaccurate. This paper presents a dead-reckoning method implementing a particle filter to estimate a set of likely map-matched hypotheses containing the correct solution with a high probability. Our method uses lane-level maps that feature dedicated attributes such as connectedness and adjacency. The vehicle position is essentially estimated by dead-reckoning sensors and lane detection using an intelligent camera. We also describe an integrity monitoring method for assessing the coherence of the set of hypotheses, using the fix of a global navigation satellite system receiver. The method provides in real-time a “Use/Don’t Use” characterization of the vehicle positioning information that is transmitted to safety functions, where integrity is fundamental. The performance of the proposed map-aided dead-reckoning method with integrity monitoring is evaluated using data acquired by an experimental car on suburban public roads. The results obtained validate the approach.

Tightly coupling GNSS pseudorange and Doppler measurements with other sensors is known to increase the accuracy and consistency of positioning information. Nowadays, high-accuracy geo-referenced lane marking maps are seen as key information sources in autonomous vehicle navigation. When an exteroceptive sensor such as a video camera or a lidar is used to detect them, lane markings provide positioning information which can be merged with GNSS data. In this paper, measurements from a forward-looking video camera are merged with raw GNSS pseudoranges and Dopplers on visible satellites. To create a localization system that provides pose estimates with high availability, dead reckoning sensors are also integrated. The data fusion problem is then formulated as sequential filtering. A reduced-order state space modeling of the observation problem is proposed to give a real-time system that is easy to implement. A Kalman filter with measured input and correlated noises is developed using a suitable error model of the GNSS pseudoranges. Our experimental results show that this tightly coupled approach performs better, in terms of accuracy and consistency, than a loosely coupled method using GNSS fixes as inputs.

This paper describes a novel road-matching method designed to support the real-time navigational function of cars for advanced systems applications in the area of driving assistance. This method provides an accurate estimation of position for a vehicle relative to a digital road map using Belief Theory and Kalman filtering. Firstly, an Extended Kalman Filter combines the DGPS and ABS sensor measurements to produce an approximation of the vehicle’s pose, which is then used to select the most likely segment from the database. The selection strategy merges several criteria based on distance, direction and velocity measurements using Belief Theory. A new observation is then built using the selected segment, and the approximate pose adjusted in a second Kalman filter estimation stage. The particular attention given to the modeling of the system showed that incrementing the state by the bias (also called absolute error) of the map significantly increases the performance of the method. Real experimental results show that this approach, if correctly initialized, is able to work over a substantial period without GPS.