A Perception-Driven Autonomous Urban Vehicle

DOI: 10.1007/978-3-642-03991-1_5

ABSTRACT This paper describes the architecture and implementation of an autonomous passenger vehicle designed to navigate using locally
perceived information in preference to potentially inaccurate or incomplete map data. The vehicle architecture was designed
to handle the original DARPA Urban Challenge requirements of perceiving and navigating a road network with segments defined
by sparse waypoints. The vehicle implementation includes many heterogeneous sensors with significant communications and computation
bandwidth to capture and process high-resolution, high-rate sensor data. The output of the comprehensive environmental sensing
subsystem is fed into a kino-dynamic motion planning algorithm to generate all vehicle motion. The requirements of driving
in lanes, three-point turns, parking, and maneuvering through obstacle fields are all generated with a unified planner. A
key aspect of the planner is its use of closed-loop simulation in a Rapidly-exploring Randomized Trees (RRT) algorithm, which
can randomly explore the space while efficiently generating smooth trajectories in a dynamic and uncertain environment. The
overall system was realized through the creation of a powerful new suite of software tools for message-passing, logging, and
visualization. These innovations provide a strong platform for future research in autonomous driving in GPS-denied and highly
dynamic environments with poor a priori information.

  • [Show abstract] [Hide abstract]
    ABSTRACT: The computationally efficient search for robust feasible paths for unmanned aerial vehicles (UAVs) in the presence of uncertainty is a challenging and interesting area of research. In uncertain environments, a “conservative” planner may be required but then there may be no feasible solution. In this paper, we use a chance constraint to limit the probability of constraint violation and extend this framework to handle uncertain dynamic obstacles. The approach requires the satisfaction of probabilistic constraints at each time step in order to guarantee probabilistic feasibility. The rapidly-exploring random tree (RRT) algorithm, which enjoys the computational benefits of a sampling-based algorithm, is used to develop a real-time probabilistically robust path planner. It incorporates the chance constraint framework to account for uncertainty within the formulation and includes a number of heuristics to improve the algorithm’s performance. Simulation results demonstrate that the proposed algorithm can be used for efficient identification and execution of probabilistically safe paths in real-time.
    Journal of Intelligent and Robotic Systems 08/2013; 71(2). · 0.83 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To support autonomous, in-water inspection of a ship hull, we propose and implement new techniques for coverage path planning over complex 3D structures. Our main contribution is a comprehensive methodology for sampling-based design of inspection routes, including an algorithm for planning, an algorithm for smoothing, and an analysis of probabilistic completeness. The latter two outcomes are the first of their kind in the area of coverage planning. Our algorithms give high-quality solutions over expansive structures, and we demonstrate this with experiments in the laboratory and on a 75 m Coast Guard cutter.
    The International Journal of Robotics Research 08/2013; 32(9-10):1048-1073. · 2.86 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: 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.
    Autonomous Robots 07/2013; 35(1). · 1.91 Impact Factor

Full-text (2 Sources)

Available from
May 31, 2014