this paper, which presents randomized, algorithmic techniques for path planning that are particular suited for problems that involve dierential constraints.

We present a randomized motion planner for robots that must avoid moving obstacles and achieve a specified goal under kinematic and dynamic constraints. The planner samples the robot's statetime space by picking control inputs at random and integrating the equations of motion. The result is a roadmap of sampled statetime points, called milestones, connected by short admissible trajectories. The planner does not precompute the roadmap as traditional probabilistic roadmap planners do; instead, for each planning query, it generates a new roadmap to find a trajectory between an initial and a goal statetime point. We prove in this paper that the probability that the planner fails to find such a trajectory when one exists quickly goes to 0, as the number of milestones grows. The planner has been implemented and tested successfully in both simulated and real environments. In the latter case, a vision module estimates obstacle motions just before planning starts; the planner is then allocated a small, fixed amount of time to compute a trajectory. If a change in the obstacle motion is detected while the robot executes the planned trajectory, the planner recomputes a trajectory on the fly. 1

Abstract not found

This paper deals with motion planning and control for mobile robots. The various components of an integrated architecture for the mobile robot Hilare pulling a trailer are presented. The nonholonomic path planner is based on an original steering method accounting for the small-time controllability of the system. Then the path is transformed into a trajectory by including the dynamical constraints of the system (bounded velocity and bounded acceleration). Finally motion control is addressed: thanks to a geometric transformation introducing a virtual robot, we show how to reduce the problem to a classical approach of trajectory tracking for a mobile robot moving only forward. Experimental results include two types of robot-trailer connection systems. Keywords---Mobile robot with trailer, Nonholonomic Path planning, Motion control. I. Introduction T HIS paper presents all the components required to devise a practical navigation system for a mobile robot pulling a trailer in a known en...

Abstract—This paper presents a steering method for a car-like vehicle providing smooth paths subjected to curvature constraints. We show how to integrate this steering method in a global motion planning scheme taking obstacles into account. The main idea of the paper is to consider the car as a 4-D system from a kinematic point of view and as a 3-D system from a geometric point of view of collision checking. The resulting planned motions are guaranteed to be collision-free and between two cusp points. Index Terms—Nonholonomic car-like robot, smooth motion planning. I.

Abstract not found

One of the approaches to collision-free nonholonomic motion planning is the "approximation method". The corresponding planners compute first a holonomic path among obstacles before approximating it by a concatenation of feasible collision-free paths. These methods are one of the rare ones which can lead to exact and complete planners. However, the performance of these planners (in terms of computation time and the complexity of the solution) is highly influenced by the "quality" of the first geometric path. In this work, we suggest a way to estimate this quality and present the first general approach leading to an improvement of the quality of the geometric path. This approach is based on local deformations of a holonomic potential field with respect to the nonholonomic constraints of the system. 1 Introduction Motion planning for nonholonomic systems amidst obstacles is a challenging research field concerning many practical robotics problems. Examples of these problems are most of t...

In this work a microscopic traffic interaction model is simulated as a distributed set of autonomous agents, each of which controls its own vehicle and responds to the actions of other agents. Both navigating along a road as well as static and dynamic obstacle avoidance are achieved by applying nonholonomic robot motion constraints on a a potential field model. Dynamic effects are controlled by constraining the turn direction to lie within a dynamically defined cone (maximum turn depends on current velocity). Each reactive agent has a set of primitive behaviours. It also has a set of beliefs regarding its own behaviour and the behaviours of other vehicles in the system. 1 Introduction Modeling vehicular interactions, difficult enough in the simplest of circumstances, is made more complex by the variability in vehicle and driver characteristics typical of uncontrolled road access. Pedestrians, trucks, automobiles, and non-motorized vehicles, share the same road area and ...

This paper surveys some recent theoretical work in incorporating the dynamics of robot models in motion planning algorithms. Such work is often referred in the literature as kinodynamic motion planning. We first give an overview of the problem in general and how it differs from motion planning with pure kinematics. We then describe three model algorithms, each representing one of three broad categories of existing methods to attack problems involving kinodynamic motion planning. For each algorithm, we review the problem considered, assumptions made, the key ideas, experimental results, and possible extensions. Computational complexity, as well as other issues relevant to practical implementations are discussed, including areas of future research. 1 Introduction Motion planning algorithms attempt to find a sequence of actions to move a system from an initial to a goal configuration [7]. The seminal paper by Schwarth and Shahir on the generalized piano movers' problem sparked a great de...

This paper presents a variant of a probabilist approach to motion planning. The objective is to face large, complex industrial environments for maintenance purpose. The contribution is based on a structuring of the workspace into boxes that limits the size of the graph to be searched. The algorithm was integrated in a commercial CAD system and real-size experiments are presented. 1 Introduction Feasibility studies for the planning of maintainance interventions at industrial sites are often done using CAD systems. Scenes have high complexity (more than 10000 polyhedral facets). Commonly devices like carts and cranes (with up to 4 DOFs, occasionally up to 8 DOFs, a tree-like kinematic structure) intervene and must carry objects through the scene. Searching for collision free motions through densely populated scenes tried by hand is a difficult and time consuming task. Automated mobility analysis therefore would considerably help to improve quality and to reduce the cost of such feasibi...