Domokos Kiss

Path planning and control of differential and car-like robots in narrow environments

This paper presents a comprehensive solution for path planning and control of two popular types of autonomous wheeled vehicles. Differentially driven and car-like motion systems are the most widespread structures among wheeled mobile robots. The planning algorithm employs a rapidly exploring random tree based global planner (RTR), which generates paths made of straight motion and in place turning primitives. Such paths can be directly followed by a differential drive robot. Carlike robots have a minimum turning radius constraint, hence we present a local steering method (C*CS) which obtains a path consisting circular and straight movements based on the primary RTR-path, without losing the existence of the solution. Additionally, a velocity profile generation algorithm is presented, which is responsible for the distribution of the time parameter along the geometric path, taking the physical limitations of the robot into account. Finally, control algorithms for path following are given for both robot types. Simulations and real experiments show the effectiveness of these methods, even is constrained environments containing narrow corridors and passages.

Advanced dynamic window based navigation approach using model predictive control

A well-known reactive motion planning technique, the dynamic window approach (DWA) provides an elegant way to navigate safely in the presence of obstacles, also taking the dynamic properties of the robot into account. Most of the DWA-based methods have the same limitation, namely they use an objective function consisting of weighted terms. Different situations require different weights, however, there is no algorithm for choosing them. This paper presents a global dynamic window-based navigation scheme using model predictive control and having no weighted objective function. Former DWA-based methods take dynamic limitations of the robot by acceleration constraints into account. In contrast with that, the proposed approach utilizes a dynamic motion model of the robot.

A receding horizon control approach to obstacle avoidance

The dynamic window approach to collision avoidance is a well-known method in the mobile robotics field. It provides an elegant way to navigate safely in the presence of obstacles, also taking the dynamic properties of the robot into account. There are different variants of this method, all having the same limitation, namely they use an objective function consisting of weighted terms. Different situations require different weights, however, there is no algorithm for choosing them. An inappropriate weighting can inhibit reaching the goal. This paper presents a navigation scheme based on the idea of dynamic window approach (DWA) and using receding horizon control (RHC). This method uses an objective function that does not contain weights. Kinematic and dynamic constraints of the robot are taken into account and a safe and goal-oriented motion is ensured.

Investigation of Dynamic Window based navigation algorithms on a real robot

The Dynamic Window Approach (DWA) to collision avoidance is a well-known and elegant method in the mobile robotics field. It provides a collision-free navigation between obstacles, and takes the dynamic properties of the robot into account. Numerous variants had been proposed for this approach, most of them are using an objective function consisting of weighted terms. The solution named Global Dynamic Window Approach with Receding Horizon Control (GDWA/RHC) uses a global navigation function - which is a scalar-valued function representing the distance from the goal point - as the objective function. In this paper, a real-world implementation of the DWA and the GDWA/RHC methods is introduced.

A Receding Horizon Control Approach to Navigation in Virtual Corridors

Applications in mobile robotics require safe and goal-oriented motion while navigating in an environment obstructed by obstacles. The dynamic window approach (DWA) to collision avoidance and its different variants provide safe motion among obstacles, although they have the same limitation, namely using an objective function consisting of weighted terms. Different situations require different weights; however, there is no algorithm for choosing them. The Global Dynamic Window Approach with Receding Horizon Control (GDWA/RHC) presented in this chapter is similar to DWA but it uses a global navigation function (NF) and a receding horizon control scheme for guiding the robot. In order to make the calculation of the navigation function computationally tractable it is constructed by interpolation from a discrete function. In addition to that the domain of the navigation function is restricted to a virtual corridor between the start and goal positions of the robot.

Autonomous Path Planning for Road Vehicles in Narrow Environments: An Efficient Continuous Curvature Approach

In this paper we introduce a novel method for obtaining good quality paths for autonomous road vehicles (e.g., cars or buses) in narrow environments. There are many traffic situations in urban scenarios where nontrivial maneuvering in narrow places is necessary. Navigating in cluttered parking lots or having to avoid obstacles blocking the way and finding a detour even in narrow streets are challenging, especially if the vehicle has large dimensions like a bus. We present a combined approximation-based approach to solve the path planning problem in such situations. Our approach consists of a global planner which generates a preliminary path consisting of straight and turning-in-place primitives and a local planner which is used to make the preliminary path feasible to car-like vehicles. The approximation methodology is well known in the literature; however, both components proposed in this paper differ from existing similar planning methods. The approximation process with the proposed local planner is proven to be convergent for any preliminary global paths. The resulting path has continuous curvature which renders our method well suited for application on real vehicles. Simulation experiments show that the proposed method outperforms similar approaches in terms of path quality in complicated planning tasks.

A Planning Method to Obtain Good Quality Paths for Autonomous Cars

Path planning among obstacles for nonholonomic systems is a widely researched area nowadays, but it is still one of the most challenging problems in autonomous navigation. We have recently presented a rapidly exploring random tree based global planner (RTR) and a steering method (C*CS) for car-like vehicles, which uses circular and straight movements. With the aid of these two methods it is possible to obtain a feasible path, even in narrow spaces and in situations requiring non-trivial maneuvers. This paper presents an improved solution for environments, where not all obstacles are known at the beginning and these are discovered during the motion of the robot. We also introduce an extension to the presented algorithm to achieve paths of better quality, i.e. more similar to a reasonable path generated by a human driver.

Effective navigation in narrow areas: A planning method for autonomous cars

Development of driverless road vehicles is one of the most active research areas of robotics today. Path planning among obstacles is one of the challenging problems to be solved in order to achieve autonomous navigation. In this paper we present a geometric path planning approach for car-like robots, intended for generating good quality paths even in cluttered environments containing narrow areas. The presented planner is designed to cope with situations which need nontrivial maneuvering between obstacles. The resulting paths are similar to those a human driver would find and have continuous curvature profile, which makes them appropriate for application on real cars. A comparative analysis of our method with possible alternatives in the literature is presented to illustrate its effectiveness regarding path quality and computation time.