Autonomous planetary rovers operating in vast unknown environments must operate efficiently because of size, power and computing limitations. Recently, we have developed a rover capable of efficient obstacle avoidance and path planning. The rover uses binocular stereo vision to sense potentially cluttered outdoor environments. Navigation is performed by a combination of several modules that each ÒvoteÓ for the next best action for the robot to execute. The key distinction of our system is that it produces globally intelligent behavior with a small computational resourceÑ all processing and decision making is done on a single processor. These algorithms have been tested on our prototype rover, Bullwinkle, outdoors and have recently driven the rover 100 m at speeds of 15 cm/ sec. In this paper we report on the extensions on the systems that we have previously developed that were necessary to achieve autonomous navigation in this domain.

The problem of path-finding in commercial computer games has to be solved in real time, often under constraints of limited memory and CPU resources. The computational effort required to find a path, using a search algorithm such as A*, increases with size of the search space. Hence, pathfinding on large maps can result in serious performance bottlenecks. This paper presents HPA * (Hierarchical Path-Finding A*), a hierarchical approach for reducing problem complexity in path-finding on grid-based maps. This technique abstracts a map into linked local clusters. At the local level, the optimal distances for crossing each cluster are pre-computed and cached. At the global level, clusters are traversed in a single big step. A hierarchy can be extended to more than two levels. Small clusters are grouped together to form larger clusters. Computing crossing distances for a large cluster uses distances computed for the smaller contained clusters. Our method is automatic and does not depend on a specific topology. Both random and real-game maps are successfully handled using no domainspecific knowledge. Our problem decomposition approach works very well in domains with a dynamically changing environment. The technique also has the advantage of simplicity and is easy to implement. If desired, more sophisticated, domain-specific algorithms can be plugged in for increased performance. The experimental results show a great reduction of the search effort. Compared to a highly-optimized A*, HPA * is shown to be up to 10 times faster, while finding paths that are within 1 % of optimal. 1 1

With the success of Mars Pathfinder's Sojourner rover, a new era of planetary exploration has opened, with demand for highly capable mobile robots. These robots must be able to traverse long distances over rough, unknown terrain autonomously, under severe resource constraints. Much prior work in mobile robot path planning has been based on assumptions that are not truly applicable to navigation through planetary terrains. Based on the author's firsthand experience with the Mars Pathfinder mission, this work reviews issues which are critical for successful autonomous navigation of planetary rovers. No current methodology addresses all of these constraints. We next develop the sensor-based "Wedgebug" motionplanning algorithm. This algorithm is complete, correct, requires minimal memory for storage of its world model, and uses only on-board sensors, which are guided by the algorithm to e#ciently sense only the data needed for motion planning, while avoiding unnecessary robot motion. The p...

Mobile robots operating in outdoor unstructured environments often have only incomplete maps and must deal with new objects found during traversal. Path planning in these environments must be incremental to accommodate new information and must use efficient representations. This article reports recent results in path planning using an efficient data structure (framedquadtrees) and an optimal algorithm (D*) to incrementally replan optimal paths. We show how the use of framed-quadtrees leads to paths that are shorter and more direct than when other representations are used. We also show the difference in performance when the robot starts with no information about the world versus when it starts with partial information about the world. Our results indicate that, as would be expected, starting with partial information is better than starting with no information. However, in many cases, partial information results in performance that is almost as good as starting out with complete information about the world, while the computational cost incurred is significantly lower. Our system has been tested in simulation as well on an autonomous jeep equipped with local obstacle avoidance capabilities. Results from both simulation and real experimentation are discussed.

This paper describes the design and implementation of a path planning system for an autonomous ground vehicle. The system is designed to be flexible, allowing any planning algorithm to be used and any topology of data to be planned over. It employs a hierarchical separation of two planning modules in conjunction with a vehicle model, to achieve continued vehicle motion while planning and the ability to act as either a deliberative or reactive planner, or a hybrid of both types. Results from both simulation and field trials are presented, and demonstrate the effectiveness of this architecture on a large outdoor ground vehicle. The contributions of this paper are twofold: a flexible planning system capable of large scale missions for autonomous vehicles; and the use of a vehicle model to determine the requirements for safe operation without slowing the vehicle, and the conditions under which this cannot be achieved.

Exploration gathers information about the unknown. This information can come in many forms, from knowledge of new terrain, to rock geology, to lifeforms. The value of these different information forms to an explorer is determined by a set of information metrics, one for each form of information, that depend on the goal of the exploration task. As explorations become more complex, increasing numbers of information metrics must be considered in order to succeed. These multiple information metrics must be considered simultaneously during exploration and often conflict with each other to compete for the finite resources of the explorer. Exploration also involves making decisions, based on the collected information, to test hypotheses and collect more information in an efficient manner. This thesis introduces a new exploration technique which actively considers how much information can be gained from taking sensor readings as well as the cost of collecting this information. The methodology can consider multiple metrics of information simultaneously --- such as finding new terrain and identifying rock type --- as it explores and these information metrics can be easily changed to perform new and different exploration tasks. The method considers the costs, such as driving, sensing and planning times, associated with collecting the information. Exploration plans are produced which maximize the utility, information gain minus exploration costs, to the exploring robot. The multiple information metric exploration planner is used to solve two exploration problems: creating traversability maps and exploring cliffs. These tasks are performed in simulation and the information gain and exploration path lengths are compared as the information metrics are changed. The multiple informat...

views and conclusions expressed in this publication are those of the author and should not be interpreted as representing the official policies, either expressed or implied, of DARPA, or the U.S. government.

We address the global trajectory planning problem of nonholonomic mobile robots in environments with static and dynamic obstacles. The global trajectory is composed of regional feasible trajectories which satisfy the dynamics of the robot kinematic model. Piecewise constant parameterization is used to construct regional feasible trajectory and steering control, and collision avoidance criterion is derived. Performance considered in the paper include robot safety, geometry-based criteria, time-based criteria, and physics-based criteria. Regional analytic trajectory solutions facilitate performance evaluation of the global trajectory. Simulations show a good performance of the planned trajectory using the proposed scheme.

Two areas of expertise required in the production of industrial and commercial robotics are motor control and obstacle navigation algorithms. This is especially true in the field of autonomous robotic vehicles, and this application will be the focus of this work. This work is divided into two parts. Part I describes the motor types and feedback devices available and the appropriate choice for a given robotics application. This is followed by a description of the control strategies available and appropriate for a variety of situations. Part II describes the vision hardware and navigation software necessary for an autonomous robotic vehicle. The conclusion discusses how the two parts are coming together in the emerging field of electric smart car technology. The content is aimed at the robotic vehicle designer. Both parts present a contribution to the field but also survey the required background material for a researcher to enter into development. The material has been made succinct and graphical wherever appropriate. (Grant Information)

Abstract—This paper presents a method for improving the runtime of an optimal heuristic path planner (A*) so that it can run repeatedly, in real-time, in a dynamic environment. This is necessary for mobile robots navigating in dynamic environments that have moving obstacles with associated costs, such as personal space around people or buffer zones around dangerous vehicles. Our approach is to modify the search space used by the A * algorithm, increasing the size of grid cells further from the robot. This approach relies on the notion that only the area closest to the robot needs to be searched carefully; areas further from the robot can be searched more coarsely. Because the planner is assumed to run repeatedly as the robot moves, the robot will always have a fine-grained path defined for its next action. We have experimentally verified in simulation that this algorithm can be run in real-time and produces paths that are comparable to full-resolution planning. I.