Georges S. Aoude

Driver Behavior Classification at Intersections and Validation on Large Naturalistic Data Set

The ability to classify driver behavior lays the foundation for more advanced driver assistance systems. In particular, improving safety at intersections has been identified as a high priority due to the large number of intersection-related fatalities. This paper focuses on developing algorithms for estimating driver behavior at road intersections and validating them on real traffic data. It introduces two classes of algorithms that can classify drivers as compliant or violating. They are based on (1) support vector machines and (2) hidden Markov models, which are two very popular machine learning approaches that have been used successfully for classification in multiple disciplines. However, existing work has not explored the benefits of applying these techniques to the problem of driver behavior classification at intersections. The developed algorithms are successfully validated using naturalistic intersection data collected in Christiansburg, VA, through the U.S. Department of Transportation Cooperative Intersection Collision Avoidance System for Violations initiative. Their performances are also compared with those of three traditional methods, and the results show significant improvements with the new algorithms.

Probabilistically safe motion planning to avoid dynamic obstacles with uncertain motion patterns

This paper presents a real-time path planning algorithm that guarantees probabilistic feasibility for autonomous robots with uncertain dynamics operating amidst one or more dynamic obstacles with uncertain motion patterns. Planning safe trajectories under such conditions requires both accurate prediction and proper integration of future obstacle behavior within the planner. Given that available observation data is limited, the motion model must provide generalizable predictions that satisfy dynamic and environmental constraints, a limitation of existing approaches. This work presents a novel solution, named RR-GP, which builds a learned motion pattern model by combining the flexibility of Gaussian processes (GP) with the efficiency of RRT-Reach, a sampling-based reachability computation. Obstacle trajectory GP predictions are conditioned on dynamically feasible paths identified from the reachability analysis, yielding more accurate predictions of future behavior. RR-GP predictions are integrated with a robust path planner, using chance-constrained RRT, to identify probabilistically feasible paths. Theoretical guarantees of probabilistic feasibility are shown for linear systems under Gaussian uncertainty; approximations for nonlinear dynamics and/or non-Gaussian uncertainty are also presented. Simulations demonstrate that, with this planner, an autonomous vehicle can safely navigate a complex environment in real-time while significantly reducing the risk of collisions with dynamic obstacles.

Two-stage path planning approach for solving multiple spacecraft reconfiguration maneuvers

The paper presents a two-stage approach for designing optimal reconfiguration maneuvers for multiple spacecraft in close proximity. These maneuvers involve well-coordinated and highly-coupled motions of the entire fleet of spacecraft while satisfying an arbitrary number of constraints. This problem is complicated by the nonlinearity of the attitude dynamics, the non-convexity of some of the constraints, and the coupling that exists in some of the constraints between the positions and attitudes of all spacecraft. While there has been significant research to solve for the translation and/or rotation trajectories for the multiple spacecraft reconfiguration problem, the approach presented in this paper is more general and on a larger scale than the problems considered previously. The essential feature of the solution approach is the separation into two stages, the first using a simplified planning approach to obtain a feasible solution, which is then significantly improved using a smoothing stage. The first stage is solved using a bidirectional Rapidly-exploring Random Tree (RRT) planner. Then the second step optimizes the trajectories by solving an optimal control problem using the Gauss pseudospectral method (GPM). Several examples are presented to demonstrate the effectiveness of the approach for designing spacecraft reconfiguration maneuvers.

Threat-aware path planning in uncertain urban environments

This paper considers the path planning problem for an autonomous vehicle in an urban environment populated with static obstacles and moving vehicles with uncertain intents. We propose a novel threat assessment module, consisting of an intention predictor and a threat assessor, which augments the host vehicles path planner with a real-time threat value representing the risks posed by the estimated intentions of other vehicles. This new threat-aware planning approach is applied to the CL-RRT path planning framework, used by the MIT team in the 2007 DARPA Grand Challenge. The strengths of this approach are demonstrated through simulation and experiments performed in the RAVEN testbed facilities.

Mobile Agent Trajectory Prediction using Bayesian Nonparametric Reachability Trees

This paper presents an efficient trajectory prediction algorithm that has been developed to improve the performance of future collision avoidance and detection systems. The main idea is to embed the inferred intention information of surrounding agents into their estimated reachability sets to obtain a probabilistic description of their future paths. More specifically, the proposed approach combines the recently developed RRT-Reach algorithm and mixtures of Gaussian Processes. RRT-Reach was introduced by the authors as an extension of the closed-loop rapidly-exploring random tree (CL-RRT) algorithm to compute reachable sets of moving objects in real-time. A mixture of Gaussian processes (GP) is a flexible nonparametric Bayesian model used to represent a distribution over trajectories and have been previously demonstrated by the authors in a UAV interception and tracking of ground vehicles planning scheme. The mixture is trained using typical maneuvers learned from statistical data, and RRT-Reach utilizes samples from the GP to grow probabilistically weighted feasible paths of the surrounding vehicles. The resulting approach, denoted as RR-GP, has RRTReach’s benefits of computing trajectories that are dynamically feasible by construction, therefore efficiently approximating the reachability set of surrounding vehicles following typical patterns. RRT-GP also features the GP mixture’s benefits of providing a probabilistic weighting on the feasible trajectories produced by RRTReach, allowing our system to systematically weight trajectories by their likelihood. A demonstrative example on a car-like vehicle illustrates the advantages of the RR-GP approach by comparing it to two other GP-based algorithms.

Sampling-Based Threat Assessment Algorithms for Intersection Collisions Involving Errant Drivers

Abstract This paper considers the decision-making problem for a vehicle crossing a road intersection in the presence of other, potentially errant, drivers. This problem is considered in a game-theoretic framework, where the errant drivers are assumed to be capable of causing intentional collisions. Our approach is to simulate the possible behaviors of errant drivers using RRT-Reach, a modified application of rapidly-exploring random trees. A novelty in RRT-Reach is the use of a dual exploration-pursuit mode, which allows for efficient approximation of the errant reachability set for some fixed time horizon. Through simulation and experimental results with a small autonomous vehicle, we demonstrate that this threat assessment algorithm can be used in real-time to minimize the risk of collision.