The task of planning trajectories for a mobile robot has received considerable attention in the research literature. Most of the work assumes the robot has a complete and accurate model of its environment before it begins to move; less attention has been paid to the problem of partially known environments. This situation occurs for an exploratory robot or one that must move to a goal location without the benefit of a floorplan or terrain map. Existing approaches plan an initial path based on known information and then modify the plan locally or replan the entire path as the robot discovers obstacles with its sensors, sacrificing optimality or computational efficiency respectively. This paper introduces a new algorithm, D*, capable of planning paths in unknown, partially known, and changing environments in an efficient, optimal, and complete manner. >

/pdf/optimal-and-efficient-path-planning-for-partially-known-4kl2zepb7v.pdf

Optimal and efficient path planning for partially-known environments

: The task of planning trajectories for a mobile robot has received considerable attention in the research literature. Algorithms exist for handling a variety of robot shapes, configurations, motion constraints, and environments. Most of the work assumes the robot has a complete and accurate model of its environment before it begins to move; less attention has been paid to the problem of unknown or partially-known environments. This situation occurs for an exploratory robot or one that must move to a goal location without the benefit of a floorplan (indoor) or terrain map (outdoor). Existing approaches plan an initial global path or route based on known information and then modify the plan locally as the robot discovers obstacles with its sensors. While this strategy works well in environments with small, sparse obstacles, it can lead to grossly suboptimal and incomplete results in cluttered spaces. An alternative approach is to replan the global path from scratch each time a new obstacle is discovered.

/pdf/optimal-and-efficient-path-planning-for-unknown-and-dynamic-26mziigexi.pdf

Optimal and Efficient Path Planning for Unknown and Dynamic Environments

Most autonomous outdoor navigation systems tested on actual robots have centered on local navigation tasks such as avoiding obstacles or following roads. Global navigation has been limited to simple wandering, path tracking, straight-line goal seeking behaviors, or executing a sequence of scripted local behaviors. These capabilities are insufficient for unstructured and unknown environments, where replanning may be needed to account for new information discovered in every sensor image. To address these problems, the authors developed a complete system that integrates local and global navigation. The local system uses a scanning laser rangefinder to detect and avoid obstacles. The global system uses an incremental path planning algorithm to optimally replan the global path for each detected obstacle. A control arbiter steers the robot to achieve the proper balance between safety and goal acquisition. This system was tested on a real robot and successfully drove it 1.4 kilometers to find a goal given no a priori map of the environment.

/pdf/a-complete-navigation-system-for-goal-acquisition-in-unknown-4dmtxgj6ef.pdf

A complete navigation system for goal acquisition in unknown environments

Racecar drivers have the ability to operate a vehicle at its friction limit without losing control. If autonomous vehicles or driver assistance systems had similar capabilities, many fatal accidents could be avoided. To advance this goal, an autonomous racing controller is designed to gain insights into vehicle control at the friction limits. A bicycle model and a ‘ g-g ’ diagram are used to mimic racecar drivers’ internal vehicle model. Lanekeeping steering feedback and wheel slip feedback controllers are used to imitate drivers making steering and throttle corrections according to the vehicle responses. Experimental results on a low friction surface demonstrate that the controller can robustly track a path while operating at the limits of tyre adhesion and provide insights for the future development of vehicle safety systems.

/pdf/autonomous-vehicle-control-at-the-limits-of-handling-bfbbj266cq.pdf

Autonomous Vehicle Control at the Limits of Handling

Intelligent agents perform concurrent tasks requiring interaction with a dynamic environment, under real-time constraints. Because an agent’s opportunities to perceive, reason, and act exceed its resources, it must determine which operations to perform and when to perform them so as to achieve its most important objectives. Accordingly, we view the problem of real-time performance as a problem in intelligent real-time control. We propose control requirements and present an agent architecture designed to address them. Key features include: parallel perception, action, and cognition processes, limitedcapacity I/O buffers with best-first retrieval and worst-first overflow, dynamic control planning, dynamic focus of attention, and a satisficing execution cycle. These features allow an agent to trade quality for speed of response under dynamic goals, resource limitations, and performance constraints. We illustrate the architecture in the Guardian system for intensive care monitoring and contrast it with alternative architectures.

/pdf/architectural-foundations-for-real-time-performance-in-4v4mdc0kvg.pdf

Architectural foundations for real-time performance in intelligent agents

ocusing primarily on system architecture, this article describes the current status of autonomous land vehicle (ALV) research at F Carnegie Mellon University’s Robotics Institute. We will (1) discuss issues concerning outdoor navigation; (2) describe our system’s perception, planning, and control components that address these issues; (3) examine Codger, the software system that integrates these components into a single system, synchronizing the dataflow between them (thereby rnaximizing parallelism); and (4) present the results of our experiments, problems uncovered in the process, and plans for addressing those problems. Carnegie Mellon’s ALV group has created an autonomous mobile robot system capable of operating in outdoor environments. Using two sensors-a color camera and a laser range finder-our system can drive a robot vehicle continuously on a network of sidewalks, up a bicycle slope, and over a curved road through an area populated with trees. The complexity of real-world domains and requirements for continuous and real-time motion require that such robot systems provide architectural support for multiple sensors and parallel processing-capabilities not found in simpler robot systems. At CMU, we are studying mobile robot system architecture and have developed a navigation system working at two test sites and on two experimental vehicles.

/pdf/mobile-robot-navigation-the-cmu-system-28ntz0cl1g.pdf

Mobile Robot Navigation: The CMU System

Mobile robot vehicles must control the execution of numerous perception and planning processes to navigate successfully in complex environments. In the past, most mobile robot systems have utilized “stop-and-go” control schemes that avoid addressing the driving control problem, or have used fixed control schemes that do not allow for the changing environment and field of view of the vehicle. This paper presents a new architecture for mobile robot control called the “Driving Pipeline”, that integrates multiple perception and planning processes and provides continuous motion with adaptive control. The Driving Pipeline has been implemented and tested on numerous versions of two vehicles: the Terregator and the Navlab. It has proven to be a flexible and powerful mechanism for building integrated software for mobile robot perception and planning.

/pdf/the-driving-pipeline-a-driving-control-scheme-for-mobile-1fna3md3o5.pdf

Yoshimasa Goto

Papers

Mobile Robot Navigation: The CMU System

The Driving Pipeline: A Driving Control Scheme for Mobile Robots