François Aioun

A Markov Decision Process-based approach for trajectory planning with clothoid tentacles

The work presented in this paper focuses on reactive local trajectory planning which plays an essential role for future autonomous vehicles. The challenge is to avoid obstacles in respect to road rules while following a global reference trajectory. The planning approach used in this work is the method of clothoid tentacles generated in the egocentered reference frame related to the vehicle. Generated tentacles in a egocentered grid represent feasible trajectories by the vehicle, and in order to choose the right one, we formulate the problem as a Markov Decision Process.

Simple and Robust Experiment Design for System Identification Using Fractional Models

This paper tackles the problems of simple and robust experiment design for system identification using elementary fractional models. It is based on a frequency domain approach and allows to determine the best sine input signal maximizing D-optimality criterion of the parameters inverse covariance matrix in different contexts? First, a single parameter (any of the parameters of the elementary fractional model) is assumed to be unknown. Next, any combination of two and then three parameters are supposed to be unknown. Finally, the problem of robust experiment design is treated when a bounded interval of the estimated parameters is known, in the same contexts.

Integrating safety distances with trajectory planning by modifying the occupancy grid for autonomous vehicle navigation

The goal of the work in this paper is to use occupancy grid in integrating safety distances with the planning strategy for autonomous vehicle navigation. The challenge is to avoid static and dynamic obstacles at high speed with respect to some specific road rules while following a global reference trajectory. Our local trajectory planning algorithm is based on the method of clothoid tentacles. It consists on generating clothoid tentacles in the egocentered reference frame related to the vehicle. Using information provided from sensors, we build an occupancy grid that we modify to take into consideration safety distances. We use this modified occupancy grid to classify each tentacle as navigable or not navigable. By formulating the problem as Markov Decision Process, only one tentacle among the navigable ones is chosen as the vehicle local reference trajectory.

Scene-aware driver state understanding in car-following behaviors

This research represents the heterogeneity in car following by a hidden variable driver state, which could change due to the drivers habit, fatigue, distraction, influence of surrounding traffic etc, resulting in the heterogeneous behaviors of such as fast or slow, strong or weak response to the same level of stimuli. A probabilistic method of driver state understanding is proposed by modeling and reasoning the heterogeneity in car-following behaviors, and the influence of surrounding traffic is addressed explicitly in addition to the leader-follower pair aiming at applications in crowded real-world traffic. Experiments are conducted by using the on-road trajectory data that were collected from motorways in Beijing, where four distinctive driver states and corresponding car-following models are learnt. With online understanding of driver state, the particular car-following model is used to predict the drivers velocity control, where results of improved accuracy are demonstrated.

Generalized driver steering model based on similarities in visual-based and optimal preview controls

The paper tackles a problem of modeling a vehicle lateral control performed by human driver. Nowadays, the modeling of driver behavior is a frequently suggested research topic and a lot of different models exist. Here, the steering control models, available in literature are presented and discussed. The similarities in different models are shown and a generalized model is proposed, merging the major different models into the same mathematical structure. Finally, the correspondence of a model is validated using real drivers data collected on a driving simulator.