Notably, a salient shortfall of most outdoor mobile robots is their lack of ability to autonomously cross roads while traveling along pedestrian sidewalks in an urban outdoor environment. If it has the ability to intuitively cross a road, the robot could then travel longer distances and more complex routes than originally possible. To this effect, the authors have been developing technologies that attempt to endow such a road-crossing function to outdoor mobile robots. In this paper, a system for road-crossing landmarks detection and localization for outdoor mobile robots is presented. We show how a robot equipped with a single monocular camera and laser range finder sensor can accurately detect, identify and localize roadcrossing landmarks such as pedestrian push button boxes, zebra crossings and pedestrian lights that the robots needs to be aware of and use in order to autonomously cross roads. In addition, experimental results and future plans are discussed.
A new method for the planning and autonomous execution of a single-trajectory, velocity-independent, parallel parking manoeuvre for autonomous vehicles is presented. The procedure commences with the identification and preselection of a smooth sigmoidal trajectory known as the Gompertz curve in parametric format. Trajectory parameters are determined in real-time during the path-planning phase using an optimization scheme in order to generate a candidate path. The optimization scheme takes into account the maximum steering angles that can be physically realized and checks the generated candidate trajectory for collisions. Thereafter, the trajectory is reparametrized to arc-length format using the cubic interpolation method and the vehicle orientation at every point of the trajectory is deduced. Following that, values of the steering angle(s) are determined. In the final step, the vehicle uses deadreckoning to follows the arc-length parametrized path in reverse in order to park itself in a single-manoeuvre. The proposed method is substantiated through both extensive simulations and real sensor data.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.